# Stack Gems - Full Catalog > Sharp, opinionated reviews of developer tools. Agent-curated, human-vetted. This document contains the complete Stack Gems catalog in markdown format. For a structured index, see /llms.txt For JSON data, use GET /api/gems --- # ● AI Agents The tools building the autonomous future: agent frameworks, MCP servers, browser automation, coding assistants, memory systems, and evaluation infrastructure. *42 gems in this category* --- # Claude Code > Terminal-first coding agent with deep repository context **Category:** AI Agents **Pricing:** paid **URL:** https://www.anthropic.com/claude-code **Added:** 2026-08-19 --- ## The Hook Claude Code lives in your terminal and rewrites your entire codebase while you watch. ## What's Good Claude Code is the coding agent that finally makes terminal-first development feel like the future. Type `claude`, describe what you want, and it reads your files, writes code, runs tests, and fixes what breaks—looping until the task is done. The context window handles entire repositories, not just single files. Multi-file refactors that would take hours happen in minutes. ## The Catch The $200/month Max plan is what serious users need—the cheaper tiers hit limits fast on large codebases. Token consumption on big repositories gets expensive. There is no autocomplete; this is for delegating tasks, not for inline suggestions while you type. ## Verdict The best terminal agent for complex multi-file work. Budget for Max tier. --- # Cursor > The AI-first code editor that understands your codebase **Category:** AI Agents **Pricing:** freemium **URL:** https://cursor.com **Added:** 2026-08-19 --- ## The Hook Cursor is what VS Code would be if Microsoft had bet everything on AI from day one. ## What's Good Cursor wraps the entire editing experience in AI. Tab completion that predicts your next edit, not just the next token. Chat that sees your whole codebase. Composer agents that handle multi-file changes while you describe what you want in plain English. Background agents keep working while you do something else. ## The Catch You are switching editors—VS Code extensions mostly work, but not all. Cursor closes, your background agents stop. The context can drift on very large refactors; you will need to guide it back. ## Verdict The best all-in-one AI coding experience. Worth the editor switch. --- # Devin > Autonomous AI engineer for defined engineering tasks **Category:** AI Agents **Pricing:** paid **URL:** https://devin.ai **Added:** 2026-08-19 --- ## The Hook Devin takes a ticket, clones the repo, writes the code, and opens the PR while you sleep. ## What's Good Devin is the most hands-off AI coding tool available. Assign a well-scoped ticket via Slack or the web interface, walk away, and come back to a pull request. It plans, writes, debugs, and deploys independently. For defined backlog work—bug fixes, test coverage, migrations—Devin delivers. ## The Catch Loud Catch: Devin is paid. The $500/month price prices out small teams. Success rates drop sharply on vague tickets; this is not a free coding agent. ## Verdict Best for clean, well-defined tickets you want to hand off entirely. --- # Stagehand > AI-native browser automation that mixes code and natural language **Category:** AI Agents **Pricing:** open-source **URL:** https://stagehand.dev **GitHub:** https://github.com/browserbase/stagehand **Added:** 2026-08-19 --- ## The Hook Stagehand lets you write Playwright scripts, then swap in natural language when the selectors break. ## What's Good Stagehand solves the hardest problem in browser automation: what happens when the UI changes. Built on Playwright, it adds AI primitives—act, extract, observe—that interpret pages like a human would. You can mix deterministic code for the stable parts and natural-language instructions for the parts that shift. ## The Catch AI calls add latency and cost. The natural-language parts are only as reliable as the underlying model. Debugging AI-driven steps is harder than debugging selectors. ## Verdict The right abstraction for browser automation that needs to survive UI changes. --- # Browser Use > Open-source framework for LLM-driven browser agents **Category:** AI Agents **Pricing:** open-source **URL:** https://browser-use.com **GitHub:** https://github.com/browser-use/browser-use **Added:** 2026-08-19 --- ## The Hook Browser Use turns any LLM into a browser agent that navigates, clicks, and extracts like a human. ## What's Good Browser Use is the open-source default for teams building AI agents that need to use the web. Point it at a URL, describe what you want, and the LLM handles navigation, form filling, and data extraction. The framework abstracts browser control into a clean API that works with multiple model providers. ## The Catch You need browser infrastructure—Browser Use is a framework, not a managed service. Token consumption scales with page complexity. Anti-bot measures will block you without stealth tooling. ## Verdict The open-source foundation for browser-based AI agents. --- # Browserbase > Managed browser infrastructure for AI agents at scale **Category:** AI Agents **Pricing:** freemium **URL:** https://browserbase.com **Added:** 2026-08-19 --- ## The Hook Browserbase runs the browsers so your agents can focus on the task. ## What's Good Browserbase handles the infrastructure AI agents need to use the web: isolated browser sessions, session recordings for debugging, browser profiles for persistent logins, and concurrency that scales. Connect with Playwright, Puppeteer, Stagehand, or Browser Use without managing your own Chromium fleet. ## The Catch Another service to pay for and depend on. Complex pricing at scale—budget carefully for high-volume workloads. You are outsourcing control over the browser layer. ## Verdict The browser infrastructure that lets you focus on the agent, not the fleet. --- # Playwright MCP > Official Microsoft MCP server for Playwright browser automation **Category:** AI Agents **Pricing:** open-source **URL:** https://github.com/microsoft/playwright-mcp **GitHub:** https://github.com/microsoft/playwright-mcp **Added:** 2026-08-19 --- ## The Hook Playwright MCP gives Claude Desktop and every MCP client real browser control. ## What's Good Microsoft’s official MCP server that exposes Playwright to any MCP-compatible tool. Claude Desktop, Cursor, Windsurf, Cline—any agent that speaks MCP can now navigate, click, and extract from the web. Accessibility snapshots provide structured DOM perception instead of pixel-based selectors. ## The Catch It is a control interface, not a complete solution. You still need browser infrastructure (local Chromium or Browserbase). No built-in session management, authentication, or scaling. ## Verdict The standard way to give MCP clients browser access. --- # Mem0 > Memory infrastructure for AI agents that actually persists **Category:** AI Agents **Pricing:** freemium **URL:** https://mem0.ai **GitHub:** https://github.com/mem0ai/mem0 **Added:** 2026-08-19 --- ## The Hook Mem0 gives your agents memory that survives across sessions, users, and deployments. ## What's Good Mem0 solves the memory problem that every agent team builds once and regrets. Automatic fact extraction, deduplication, conflict resolution, and semantic retrieval in a single API. Memory scoped by user, session, agent, and application—no data mixing between tenants. ## The Catch External dependency for a core capability. Async processing adds latency. Platform optimizations are not in the open-source SDK. You are trusting Mem0 with potentially sensitive user data. ## Verdict The memory layer most agent teams should use instead of building. --- # Braintrust > Eval-first LLM platform that blocks bad prompts in CI **Category:** AI Agents **Pricing:** freemium **URL:** https://braintrust.dev **Added:** 2026-08-19 --- ## The Hook Braintrust blocks your merge when the prompt makes the model worse. ## What's Good Braintrust is built around one loop: change a prompt, measure the result, ship the better one. The Eval() primitive runs experiments with real scorers—exact match, embedding similarity, LLM-as-judge factuality—and the CI integration blocks merges on statistically significant regressions. ## The Catch Eval-first means observability is secondary—if your primary loop is debugging traces, look elsewhere. Learning curve to build eval datasets that actually represent production. ## Verdict The platform for teams whose deployment loop is change-measure-ship. --- # Langfuse > Open-source LLM observability you can self-host **Category:** AI Agents **Pricing:** open-source **URL:** https://langfuse.com **GitHub:** https://github.com/langfuse/langfuse **Added:** 2026-08-19 --- ## The Hook Langfuse is the observability platform you can run on your own infrastructure. ## What's Good Langfuse is the open-source default for LLM observability. The entire stack—web app, worker, Postgres, ClickHouse, Redis—runs on your infrastructure under MIT license. OpenTelemetry-style tracing captures the full tree of spans and generations with token counts, latency, cost, and model metadata. ## The Catch Self-hosting means self-operating—you are running Postgres, ClickHouse, Redis, and the app. Observability-first means evals are layered on, not central. ## Verdict The LLM observability platform for teams who want to own their data. --- # Aider > AI pair programming in your terminal with any model **Category:** AI Agents **Pricing:** open-source **URL:** https://aider.chat **GitHub:** https://github.com/paul-gauthier/aider **Added:** 2026-08-19 --- ## The Hook Aider is the open-source terminal coding assistant you run with your own API keys. ## What's Good Aider gives you Claude Code-style terminal coding without the subscription. Bring your own API keys from Anthropic, OpenAI, or dozens of other providers. Git-aware editing means changes are tracked automatically. Model-agnostic design means you pick the price-performance tradeoff. ## The Catch You are paying per-token at API rates—can exceed subscription costs on heavy use. No managed infrastructure; you are the ops team. Less polished than commercial alternatives. ## Verdict The open-source terminal coding assistant for API-key purists. --- # Cline > VS Code extension for human-in-the-loop AI coding **Category:** AI Agents **Pricing:** open-source **URL:** https://cline.bot **GitHub:** https://github.com/cline/cline **Added:** 2026-08-19 --- ## The Hook Cline asks before every action, making it the safest agentic coding tool available. ## What's Good Cline is the VS Code extension for developers who want AI assistance with explicit control. Every file edit, every shell command, every action requires your approval. This makes it the safest option for production code—no surprise changes, no autonomous execution you did not sanction. ## The Catch Human-in-the-loop means constant interruptions. Large tasks require many approvals. The pace is slower than autonomous alternatives. ## Verdict Maximum control for developers who approve everything. --- # Letta > Episodic memory for agents that think across sessions **Category:** AI Agents **Pricing:** open-source **URL:** https://letta.com **GitHub:** https://github.com/letta-ai/letta **Added:** 2026-08-19 --- ## The Hook Letta gives agents memory that pages in and out like an operating system. ## What's Good Letta (formerly MemGPT) pioneered the operating-system approach to agent memory. Instead of stuffing everything into context, it pages older information in and out of the working window, letting agents maintain coherent threads across sessions that would otherwise exceed context limits. ## The Catch More complex to reason about than simple vector retrieval. The paging mechanism adds latency and orchestration overhead. Documentation assumes understanding of the original MemGPT paper. ## Verdict The memory architecture for agents that need to think long-term. --- # Vercel AI SDK > Unified TypeScript SDK for building AI applications **Category:** AI Agents **Pricing:** open-source **URL:** https://sdk.vercel.ai **GitHub:** https://github.com/vercel/ai **Added:** 2026-08-19 --- ## The Hook Vercel AI SDK is the abstraction layer that makes model providers interchangeable. ## What's Good One SDK, every model provider. Streaming responses, tool calling, structured outputs—the AI SDK handles the details while you write against a unified interface. Swap between OpenAI, Anthropic, Google, and more without changing application code. ## The Catch Vercel-optimized, which is great on Vercel and requires more work elsewhere. Abstraction means some provider-specific features are harder to access. ## Verdict The SDK for AI applications in the Next.js ecosystem. --- # LangChain > The OG LLM orchestration framework **Category:** AI Agents **Pricing:** open-source **URL:** https://langchain.com **GitHub:** https://github.com/langchain-ai/langchain **Added:** 2026-08-18 --- ## The Hook LangChain is the Swiss Army knife of LLM development. ## What's Good LangChain excels at chaining together prompts, tools, and retrievers into coherent workflows. LCEL made composition feel natural after years of callback hell. The integration ecosystem is unmatched: if a vector database or LLM provider exists, LangChain probably has a connector. ## The Catch Abstraction layers add complexity. Simple use cases feel over-engineered, and the learning curve is steeper than rolling your own for basic RAG. ## Verdict Essential for complex LLM apps, overkill for simple ones. --- # LangGraph > Stateful multi-actor agent workflows **Category:** AI Agents **Pricing:** open-source **URL:** https://langchain-ai.github.io/langgraph/ **GitHub:** https://github.com/langchain-ai/langgraph **Added:** 2026-08-18 --- ## The Hook LangGraph is the operating table for agents that need to loop, branch, and remember. ## What's Good The graph-based approach finally makes agentic behavior tractable. You define nodes (actions), edges (transitions), and state schemas—then the framework handles persistence, streaming, and human-in-the-loop interrupts. ## The Catch The abstraction overhead is real. Debugging graph execution requires understanding the underlying state management, and simple agents become verbose. ## Verdict The right tool when agents need memory and branching logic. --- # CrewAI > Role-playing agents that actually ship **Category:** AI Agents **Pricing:** open-source **URL:** https://crewai.com **GitHub:** https://github.com/crewAIInc/crewAI **Added:** 2026-08-18 --- ## The Hook CrewAI makes multi-agent collaboration as simple as defining who does what. ## What's Good The mental model is delightfully simple: define agents with roles, goals, and backstories, then let them work together on tasks. The role-playing paradigm sounds gimmicky but actually works—agents with defined personas produce more consistent outputs. ## The Catch The simplicity cuts both ways. Complex inter-agent communication patterns require workarounds, and you are locked into CrewAI execution model. ## Verdict Best DX for multi-agent systems if you embrace the paradigm. --- # Haystack > Production NLP pipelines from deepset **Category:** AI Agents **Pricing:** open-source **URL:** https://haystack.deepset.ai **GitHub:** https://github.com/deepset-ai/haystack **Added:** 2026-08-18 --- ## The Hook Haystack was doing RAG before RAG was cool, and it shows. ## What's Good The pipeline abstraction is elegant: components connect through well-defined interfaces. Haystack 2.0 rewrite made composition more intuitive without sacrificing power. The evaluation framework is a standout. ## The Catch Haystack strength is also its constraint. The framework assumes pipeline-based architectures; more exotic agent patterns require workarounds. ## Verdict The mature choice for document-centric AI applications. --- # LlamaIndex > Data framework for LLM applications **Category:** AI Agents **Pricing:** open-source **URL:** https://llamaindex.ai **GitHub:** https://github.com/run-llama/llama_index **Added:** 2026-08-18 --- ## The Hook LlamaIndex connects LLMs to data sources that are not just documents. ## What's Good The connector ecosystem is unmatched: 160+ data loaders handle everything from Notion to Snowflake. The indexing abstractions (vector, keyword, knowledge graph) let you pick the right retrieval strategy for your data shape. ## The Catch The abstraction layers can be confusing. Index, query engine, retriever, synthesizer—the terminology requires investment to master. ## Verdict The best data plumbing for LLM applications. --- # OpenAI Codex CLI > Terminal coding agent from OpenAI **Category:** AI Agents **Pricing:** freemium **URL:** https://github.com/openai/codex **GitHub:** https://github.com/openai/codex **Added:** 2026-08-19 --- ## The Hook OpenAI Codex CLI is the terminal rival to Claude Code, backed by OpenAI infrastructure. ## What's Good OpenAI enters the terminal agent race with their own coding CLI. The open-source client connects to OpenAI models with the same agentic loop as Claude Code—read files, write code, run commands, iterate. Deep integration with OpenAI ecosystem and models. The CLI itself is open; you can read the code and understand how it works. ## The Catch The CLI is open-source, but the frontier models powering it sit behind ChatGPT Plus, Pro, or Enterprise subscriptions. You are paying OpenAI either way. Feature parity with Claude Code is still catching up. ## Verdict OpenAI answer to terminal agents. Open CLI, paid models. --- # Goose > Local-first Rust agent with MCP at its core **Category:** AI Agents **Pricing:** open-source **URL:** https://goose-docs.ai **Added:** 2026-08-19 --- ## The Hook Goose is the local-first, MCP-native agent from the Linux Foundation AAIF. ## What's Good Goose runs locally on your machine, written in Rust for performance. MCP is a first-class citizen, not an afterthought—tools connect via the protocol natively. Part of the Linux Foundation AI and Data Foundation (AAIF), giving it institutional backing without corporate lock-in. Bring your own model provider. ## The Catch General-purpose agent means you configure it for your use case. You bring the model—Goose is the runtime, not the intelligence. Documentation is still maturing. ## Verdict Local-first, MCP-native, model-agnostic. Bring your own brain. --- # OpenHands > Self-hostable autonomous software engineer **Category:** AI Agents **Pricing:** open-source **URL:** https://openhands.dev **Added:** 2026-08-19 --- ## The Hook OpenHands is the open-source, self-hostable counterpart to Devin. ## What's Good Full autonomy like Devin—give it a task, let it plan, code, debug, and deliver—but you own the infrastructure. Self-hosting means no per-task fees and complete control over your code. Active development community. Works with multiple model providers. ## The Catch Heavier than lighter agents like Goose. Self-hosting autonomous agents requires infrastructure and monitoring. The autonomy that makes it powerful also makes debugging failures harder. ## Verdict Self-hosted Devin alternative for teams who want control. --- # OpenClaw 2.0 > Coding and personal agent across 29 messaging channels **Category:** AI Agents **Pricing:** freemium **URL:** https://openclaw.ai **Added:** 2026-08-19 --- ## The Hook OpenClaw 2.0 brings coding agents to WhatsApp, Telegram, Slack, and 26 other channels. ## What's Good Meet users where they already are. OpenClaw runs as a coding and personal assistant across 29 messaging platforms—WhatsApp, Telegram, Slack, Discord, and more. No new app to install; just message. The multi-channel approach makes AI assistance genuinely accessible. ## The Catch Huge security surface. Every connected channel is a potential attack vector. You are trusting OpenClaw with access to your messaging platforms and the code it writes. Evaluate carefully for sensitive work. ## Verdict AI assistant in your existing chat apps. Mind the security surface. --- # Needle 2 > 14MB on-device tool-calling model **Category:** AI Agents **Pricing:** open-source **URL:** https://github.com/cactus-compute/needle **GitHub:** https://github.com/cactus-compute/needle **Added:** 2026-08-19 --- ## The Hook Needle 2 is a 14MB model that runs tool-calling on-device. ## What's Good Cactus Compute built a tool-calling model that fits in 14MB. Runs entirely on-device—no API calls, no latency, no costs per call. Purpose-built for structured function calling, not general chat. The small footprint makes it embeddable anywhere. ## The Catch Not a chat LLM. Needle does one thing: parse user intent into tool calls. You still need the tools and the orchestration around it. Do not expect conversation or reasoning. ## Verdict Tiny on-device tool-caller. One job, done locally. --- # Mastra > TypeScript agent runtime with typed agents and workflows **Category:** AI Agents **Pricing:** open-source **URL:** https://mastra.ai **Added:** 2026-08-19 --- ## The Hook Mastra is the TypeScript runtime for building typed agents, workflows, and evals. ## What's Good Type-safe agent development in TypeScript. Define agents with proper types, compose them into workflows, run evals to measure quality. The framework approach means structure and reusability. Good fit for teams building agent systems rather than one-off scripts. ## The Catch Framework, not a coding CLI. Mastra helps you build agents; it is not an agent itself. Learning curve for the runtime concepts. You are committing to their abstractions. ## Verdict TypeScript agent framework for teams building agent systems. --- # E2B > Firecracker sandboxes for untrusted agent code **Category:** AI Agents **Pricing:** freemium **URL:** https://e2b.dev **Added:** 2026-08-19 --- ## The Hook E2B gives agents a safe place to run untrusted code. ## What's Good When your agent writes code, where does it run? E2B provides Firecracker-based sandboxes—lightweight VMs that spin up in milliseconds, execute untrusted code, and tear down cleanly. The infrastructure lets agents be dangerous without risking your systems. SDKs for Python and TypeScript. ## The Catch Hosted by default. You are sending code to E2B infrastructure. Self-hosting is possible but requires more work. Another external dependency in your agent stack. ## Verdict Safe sandboxes for agent-generated code. Hosted convenience. --- # Gemini CLI > Google open-source terminal agent powered by Gemini **Category:** AI Agents **Pricing:** open-source **URL:** https://geminicli.com **GitHub:** https://github.com/google-gemini/gemini-cli **Added:** 2026-09-03 --- ## The Hook Gemini CLI is Google open-source answer to terminal coding agents. ## What's Good Install the @google/gemini-cli package globally for a terminal agent backed by Gemini 3 with a 1M-token context window. Built-in Google Search grounding, file ops, command execution, web fetch, and MCP extensions. Apache 2.0, GitHub Action for PR review and issue triage, headless JSON mode for scripts. ## The Catch Gemini-bound lock-in—not model-agnostic like Goose or Crush. Consumer cutoff 2026-06-18: free/Pro/Ultra access dies; enterprise and API keys stay. Do not bet a workflow on the old Google-login quota path. The OSS CLI remains. Antigravity CLI is the successor, not a 1:1 swap. Preview/nightly channels move fast; pin stable for production workflows. Enterprise path pushes you toward Vertex AI and Google Cloud project setup. ## Verdict Google terminal agent. OSS CLI remains, but consumer quota was cut 2026-06-18; Antigravity (agy) is the consumer path. Gemini-bound by design. --- # Chrome DevTools MCP > Official Chrome DevTools as an MCP server for coding agents **Category:** AI Agents **Pricing:** open-source **URL:** https://github.com/ChromeDevTools/chrome-devtools-mcp **GitHub:** https://github.com/ChromeDevTools/chrome-devtools-mcp **Added:** 2026-09-03 --- ## The Hook Chrome DevTools MCP gives Cursor, Claude, and Copilot a real Chrome browser with DevTools attached. ## What's Good Official Chrome DevTools team MCP server. Agents get performance traces, network inspection, console with source-mapped stacks, screenshots, and Puppeteer-backed automation that waits for results. Works with Antigravity, Claude, Cursor, Copilot. Slim mode for basic browsing; full mode for debugging. This is the browser surface built for coding agents, not a generic scraper. ## The Catch chrome-devtools-mcp@1.8.0 engines: "^20.19.0 || ^22.12.0 || >=23". 20.16 and 20.17 fail; 20.19 LTS is in engines. “Node ≥ 22.12” and “Node 20 fails” are both too blunt — README “LTS” is marketing; engines is the catch. A CLI is provided for use without MCP; official CLI docs call it experimental. Complementary to Playwright MCP already on the catalog—DevTools-grade traces, Lighthouse, heap snapshots, and --autoConnect, not a generic scraper or a Playwright replacement. Officially supports Google Chrome and Chrome for Testing only. Exposes browser content to the MCP client—do not point it at sensitive sessions. Usage statistics on by default (opt out with --no-usage-statistics). Needs a current Chrome. ## Verdict The MCP browser for agents that need real DevTools, not just clicks. --- # Crush > Charm-grade TUI coding agent with LSP and mid-session model switch **Category:** AI Agents **Pricing:** source-available **URL:** https://github.com/charmbracelet/crush **GitHub:** https://github.com/charmbracelet/crush **Added:** 2026-09-03 --- ## The Hook Crush is the terminal coding agent that looks and feels like a Charm app. ## What's Good Multi-model from day one—switch LLMs mid-session without losing context. LSP-enhanced context, MCP over http/stdio/sse, Agent Skills support, and sessions per project. Runs on macOS, Linux, Windows, Android, FreeBSD, and more. Hyper (Charm) is the optional freemium provider; bring Anthropic, OpenAI, Gemini, OpenRouter, Ollama, Bedrock, Vertex, or custom OpenAI-compat endpoints. Built by the team behind Glow, Bubble Tea, and Lip Gloss. ## The Catch Licensed FSL-1.1-MIT (competing-use restriction; source-available; MIT after two years—not OSI open-source). Pre-1.0 energy—config is migrating from JSON to crushrc Bash. Not a fully autonomous SWE agent like OpenHands; you stay in the loop. Hyper is Charm own model path if you want a subscription. Metrics on by default (DO_NOT_TRACK / CRUSH_DISABLE_METRICS to opt out). ## Verdict The beautiful, model-flexible TUI coding agent. Charm quality. --- # Pydantic AI > Typed Python agent loop with a snap-on Coder harness **Category:** AI Agents **Pricing:** open-source **URL:** https://ai.pydantic.dev **GitHub:** https://github.com/pydantic/pydantic-ai **Added:** 2026-09-03 --- ## The Hook Pydantic AI is how the Pydantic team does agents—typed end to end, not another wrapper around the validator. ## What's Good This is a different product from Pydantic-the-validator already on the Python stack. Core is a typed agent loop: structured output, typed deps, typed tools, model as a string swap (OpenAI, Anthropic, Gemini, Bedrock, Groq, Ollama, and more). Pydantic AI Harness snaps on Coder—workspace-rooted files, allowlisted shell, repo orientation, planning, sub-agents, context management—as a regular combined capability you can use whole or take apart (FileSystem, Shell, RepoContext, Planning). Same agent runs as CLI (clai), web chat, realtime voice, or a durable Temporal/DBOS/Prefect job. MIT. ## The Catch ACP (Agent Client Protocol, for editors like Zed) is experimental: it lives under pydantic_ai_harness.experimental.acp, may change or be removed in any release without a deprecation period, and importing it emits HarnessExperimentalWarning. Coder is a capability you assemble, not a turnkey SWE agent like OpenHands. Logfire is the convenient observability path; OpenTelemetry works with any backend. Do not confuse this gem with Pydantic validation. ## Verdict The typed Python agent SDK. Not the validator. ACP is still experimental. --- # Worktrunk > Git worktrees that spawn an agent in one switch **Category:** AI Agents **Pricing:** open-source **URL:** https://worktrunk.dev **GitHub:** https://github.com/max-sixty/worktrunk **Added:** 2026-09-03 --- ## The Hook `wt switch -c -x claude feat` creates the worktree and starts the agent in it. ## What's Good Worktrunk is a Git worktree CLI built for parallel agent sessions. Address worktrees by branch name; `wt switch -c -x claude feat` is create + cd + start Claude instead of `git worktree add` plus a manual launch. Hooks run at switch/create/merge/remove. `wt step copy-ignored` in post-start copies gitignored caches—`node_modules/`, `target/`—so a new tree does not cold-start. Dual-licensed MIT OR Apache-2.0. ## The Catch On Windows, `wt` is Windows Terminal's command. The installer ships `git-wt` to avoid the collision (`winget install max-sixty.worktrunk` then `git-wt config shell install`); you can also disable the Terminal app-execution alias. Project hooks require first-run approval. This is worktree plumbing, not a coding agent. ## Verdict The worktree CLI for running several agents at once. On Windows, call git-wt. --- # agent-browser > Vercel Labs Rust CLI with compact a11y refs for agents **Category:** AI Agents **Pricing:** open-source **URL:** https://agent-browser.dev **GitHub:** https://github.com/vercel-labs/agent-browser **Added:** 2026-09-03 --- ## The Hook agent-browser snapshots a compact accessibility tree and lets the agent click `@e1` instead of a CSS selector. ## What's Good Vercel Labs Rust CLI (Apache-2.0). Native daemon talks CDP to Chrome. `agent-browser snapshot -i` returns an a11y tree with refs (`@e1`, `@e2`); click/fill use those refs. Distinct from Browser Use (Python LLM browser framework), Playwright MCP (MCP control plane), and Chrome DevTools MCP (DevTools traces/Lighthouse). CLI-first for any agent that can run a shell—Claude Code, Cursor, Codex, Gemini. Native binaries for macOS, Linux, Windows. ## The Catch Default idle timeout is about one hour: the headless daemon shuts down and discards transient tabs/state unless you passed `--restore` (or another restore key). Headed, Safari/iOS WebDriver, and user-attached browsers are exempt; cloud-provider browsers are not. Tune with `--idle-timeout` or `AGENT_BROWSER_IDLE_TIMEOUT_MS`; `0` disables. This is CDP Chrome, not stealth—anti-bot pages will still block you. ## Verdict The CLI browser for agents that want refs, not a full DOM dump. Restore or lose the session. --- # Zed > GPL-3.0 GPU-accelerated editor with first-party agents **Category:** AI Agents **Pricing:** open-source **URL:** https://zed.dev **GitHub:** https://github.com/zed-industries/zed **Added:** 2026-09-03 --- ## The Hook Zed is the Atom team's next editor — fast, collaborative, and licensed GPL-3.0. ## What's Good GPL-3.0 (LICENSE-GPL). Native macOS, Linux, and Windows. Multiplayer editing, language servers, first-party agent features in the editor. Download from zed.dev/download or a package manager. ## The Catch GPL-3.0 is copyleft, not MIT. Hosted agent features that talk to Zed's models are a product on top of the editor — this review does not invent a Cloud free tier. Web is not shipped. ## Verdict The fast open editor. GPL-3.0, agents in-process, no invented Cloud giveaway. --- # Hatchet > MIT durable tasks/agents on Postgres; Cloud free tier real **Category:** AI Agents **Pricing:** freemium **URL:** https://hatchet.run **GitHub:** https://github.com/hatchet-dev/hatchet **Added:** 2026-09-04 --- ## The Hook Hatchet is durable workflow/agent orchestration on the Postgres you already run when Temporal is the heavier cluster. ## What's Good MIT engine. Queues, retries, durable workflows, DAG-ish pipelines, Python/TS/Go/Ruby SDKs. Hatchet Lite single image for low throughput; Compose + RabbitMQ for prod. Cloud: first 100k task runs/mo free (Developer). ## The Catch Vs Temporal/Inngest/Trigger.dev: Postgres-backed durable tasks + agent runs — not Temporal’s own cluster story, not Inngest/Trigger event-function DX alone. Cloud paid past free run allotment. Self-host is $0 software. ## Verdict MIT durable tasks on Postgres. Cloud free tier real; self-host forever. --- # LiteLLM > MIT OpenAI-compatible gateway to 100+ LLMs; Enterprise SSO paid **Category:** AI Agents **Pricing:** freemium **URL:** https://www.litellm.ai **GitHub:** https://github.com/BerriAI/litellm **Added:** 2026-09-04 --- ## The Hook LiteLLM is the self-host OpenAI-format proxy when every agent SDK speaking a different provider API is the mess. ## What's Good MIT core SDK + proxy. One OpenAI-compatible API across 100+ providers. Virtual keys, spend tracking, budgets, fallbacks, load balancing. Docker/Helm self-host; no telemetry required when you run it yourself. ## The Catch Vs OpenRouter/provider SDKs: self-host gateway you operate — not a hosted model marketplace. LOUD: enterprise/ tree and SSO/audit/RBAC are commercial; core proxy stays MIT. You still pay model providers. ## Verdict MIT LLM gateway. Self-host free; Enterprise SSO/governance is paid. --- # Windmill > AGPLv3 scripts→workflows/UIs/agents; CE free self-host unlimited runs **Category:** AI Agents **Pricing:** freemium **URL:** https://www.windmill.dev **GitHub:** https://github.com/windmill-labs/windmill **Added:** 2026-09-04 --- ## The Hook Windmill is scripts-as-jobs with auto UIs and agent steps when Temporal is the heavy cluster and n8n is the fair-code canvas. ## What's Good AGPLv3 core. Turn TS/Python/Go/Bash/SQL into webhooks, schedules, DAG flows, and apps. CE self-host: unlimited executions, 3 workspaces, 50 users. Fast Rust job engine; MCP/agent tooling in flows. Cloud also has a free OSS tier. ## The Catch Vs Temporal/Hatchet/n8n: script+UI internal platform, not Temporal durable SDK alone or n8n canvas. LOUD: CE caps (4 groups, 10 SSO users, no durable audit logs, limited Git sync); Enterprise from ~$120/mo + seats/compute. Binary CE includes proprietary bits; from-source without enterprise flag is AGPL. ## Verdict AGPLv3 workflow/agent platform. CE self-host free with caps; EE for audit/SAML/scale. --- # Promptfoo > MIT LLM evals & red-teaming CLI; Community free, 10k probes/mo **Category:** AI Agents **Pricing:** freemium **URL:** https://www.promptfoo.dev **GitHub:** https://github.com/promptfoo/promptfoo **Added:** 2026-09-04 --- ## The Hook Promptfoo is declarative LLM evals and red-teaming when Braintrust is the hosted eval platform and you want YAML-in-CI first. ## What's Good MIT. Compare prompts/models, score outputs, CI gates, red-team/OWASP-style scans. Runs locally — prompts stay on your machine. Works with OpenAI, Anthropic, Azure, Bedrock, Ollama, custom providers. Now part of OpenAI with stated MIT permanence. ## The Catch Vs DeepEval/Braintrust: YAML/CLI-first, not pytest-native Python or Braintrust SaaS. LOUD: Community caps red-team probes (~10k/mo); Enterprise/On-Prem are contact-sales (SSO, dashboards, dedicated runner). LLM API costs are yours. Acquisition governance is a multi-year bet. ## Verdict MIT eval/red-team CLI. Free core; probe cap + Enterprise sales wall. --- # DeepEval > Apache-2.0 pytest-style LLM/agent evals; Confident AI is the paid cloud **Category:** AI Agents **Pricing:** open-source **URL:** https://deepeval.com **GitHub:** https://github.com/confident-ai/deepeval **Added:** 2026-09-04 --- ## The Hook DeepEval is pytest for LLM apps when Promptfoo is YAML red-team and Langfuse is the tracing cousin. ## What's Good Apache-2.0. 50+ metrics (G-Eval, hallucination, relevancy, task completion, multi-turn). Local-first unit/regression tests, synthetic data, CI. No paywalled metrics in the OSS package. ## The Catch OSS is the runner — collaboration/dashboards/monitoring are Confident AI (Free: 2 seats/1 project/5 runs/wk; Starter $200/mo; Team $2k/mo). LLM-as-judge burns provider tokens. Vs Ragas: broader agent/chat coverage, not RAG-only. ## Verdict Apache-2.0 LLM unit tests. Free locally; Confident AI starts free then $200. --- # Graphiti > Apache-2.0 temporal knowledge graphs for agent memory; Zep's OSS engine **Category:** AI Agents **Pricing:** open-source **URL:** https://help.getzep.com/graphiti **GitHub:** https://github.com/getzep/graphiti **Added:** 2026-09-04 --- ## The Hook Graphiti is bi-temporal agent memory as a graph when Mem0 is vector-first recall and Letta is self-editing agent state. ## What's Good Apache-2.0. Episodes → entities/edges with validity windows; hybrid retrieve (embeddings + BM25 + graph). Neo4j/FalkorDB/Kuzu/Neptune. MCP server for agent IDEs. Powers Zep's managed context lake. ## The Catch Framework only — you run a graph DB + ops. Zep Community Edition deprecated; full Zep SaaS is separate (Free credits then Flex ~$104/mo billed annually). Not a drop-in chat history store. Schema/ops cost vs Mem0 Docker. ## Verdict Apache-2.0 temporal memory graphs. Free self-host; Zep Cloud is the paid path. --- # smolagents > Apache-2.0 minimal code-agent framework from Hugging Face **Category:** AI Agents **Pricing:** open-source **URL:** https://huggingface.co/docs/smolagents **GitHub:** https://github.com/huggingface/smolagents **Added:** 2026-09-04 --- ## The Hook smolagents is the few-thousand-line agent kit when LangGraph is the heavy graph runtime and you want agents that write Python to call tools. ## What's Good Apache-2.0. CodeAgent (executable Python) + ToolCallingAgent (JSON). Hub-native tools, multi-agent orchestration, optional E2B sandbox. First-class HF model routing. Tiny core you can read. ## The Catch Minimalist — not Temporal durable workflows or production multi-tenant agent platform. Code execution needs sandbox discipline. Vs CrewAI/PydanticAI: HF Hub gravity, less enterprise packaging. You own evals, memory, and deploy. ## Verdict Apache-2.0 code-first agents. Free library; sandbox/ops are yours. --- # Ollama > MIT local model runner — free forever; Pro cloud $20/mo **Category:** AI Agents **Pricing:** freemium **URL:** https://ollama.com **GitHub:** https://github.com/ollama/ollama **Added:** 2026-09-05 --- ## The Hook Ollama is the local free-forever model runner when you want models on your machine without a coding-agent shell. ## What's Good MIT. Pull and run open models locally with a simple CLI/API. Free forever on device; optional Pro cloud at $20/mo. Huge community default for laptop LLM experiments. ## The Catch Model runner, not a coding agent. Distinct from vLLM (server serving) and LM Studio (GUI desktop app). You still pick models, prompts, and the agent layer on top. ## Verdict MIT local runner. Free on device; Pro is optional cloud. --- # Continue > Apache-2.0 open-source coding assistant — GitHub is source of truth **Category:** AI Agents **Pricing:** open-source **URL:** https://continue.dev **GitHub:** https://github.com/continuedev/continue **Added:** 2026-09-05 --- ## The Hook Continue is the open-source coding assistant whose GitHub remains the source of truth after the Cursor acquisition. ## What's Good Apache-2.0. IDE extension lineage, configurable models, OSS stays. Loud Catch is the acquisition story — product site reads memorial; repo is where the project lives. ## The Catch Cursor acquired Continue; product site is memorial, GitHub is source of truth; OSS remains. Not Cursor-the-editor. Expect docs/site lag vs the repo. ## Verdict Apache-2.0 coding assistant. Free OSS; site is memorial, repo is real. --- # ▲ Next.js The ecosystem that ships: databases, auth, APIs, and infrastructure for production React applications. *42 gems in this category* --- # Vercel > The platform that made Next.js deployment trivial **Category:** Next.js **Pricing:** freemium **URL:** https://vercel.com **Added:** 2026-08-18 --- ## The Hook Vercel made deployment someone else problem, and charges accordingly. ## What's Good Preview deployments for every PR changed how teams review frontend work. Edge functions, image optimization, and analytics are built-in. The DX improvements compound over time. Vercel investment in Next.js means new framework features work perfectly on day one. ## The Catch Vendor lock-in is real, and costs scale with success. Bandwidth and serverless function invocations add up quickly for high-traffic apps. ## Verdict Worth the premium if you ship fast and iterate often. --- # Prisma > Type-safe database access with killer DX **Category:** Next.js **Pricing:** freemium **URL:** https://prisma.io **GitHub:** https://github.com/prisma/prisma **Added:** 2026-08-18 --- ## The Hook Prisma autocomplete is addictive—once you have experienced it, raw SQL feels medieval. ## What's Good Prisma Client autocomplete is genuinely transformative for DX. Migrations work reliably, the introspection workflow handles existing databases, and Prisma Studio provides a usable data browser. ## The Catch Abstraction costs. Complex queries sometimes require raw SQL, and the generated client adds bundle size. Performance-sensitive applications need careful query optimization. ## Verdict The DX is addictive; just watch your query patterns. --- # Drizzle ORM > TypeScript ORM that feels like SQL **Category:** Next.js **Pricing:** open-source **URL:** https://orm.drizzle.team **GitHub:** https://github.com/drizzle-team/drizzle-orm **Added:** 2026-08-18 --- ## The Hook Drizzle is type-safe SQL for developers who think Prisma is too much magic. ## What's Good The query builder stays close to SQL syntax—if you know SQL, you know Drizzle. Zero dependencies and excellent tree-shaking make it edge-friendly. Migrations are SQL files, which some teams prefer. ## The Catch Less magic means more typing. Relation handling requires more manual work than Prisma. The ecosystem is smaller. ## Verdict SQL-first type safety for the abstraction-allergic. --- # tRPC > End-to-end typesafe APIs without codegen **Category:** Next.js **Pricing:** open-source **URL:** https://trpc.io **GitHub:** https://github.com/trpc/trpc **Added:** 2026-08-18 --- ## The Hook tRPC eliminates the API contract layer you did not want to maintain anyway. ## What's Good The zero-codegen approach means types flow naturally from server to client. React Query integration handles caching, mutations, and optimistic updates. The procedure-based API encourages small, focused endpoints. ## The Catch TypeScript only—this is not for mixed-language teams. The tight coupling means you lose the API as a standalone artifact. ## Verdict TypeScript APIs as they should be—type-safe by default. --- # Clerk > Auth that actually ships with your UI **Category:** Next.js **Pricing:** freemium **URL:** https://clerk.com **Added:** 2026-08-18 --- ## The Hook Clerk solves auth completely—not just the backend, but the actual forms. ## What's Good Integration with Next.js is exceptional: middleware handles route protection, hooks provide user state, and Server Components work seamlessly. Multi-factor authentication, social logins, and organization management are included. ## The Catch Loud Catch: Clerk is a paid auth product with a free tier that evaporates as MAUs grow. You are outsourcing login UI and session logic; customization is the Clerk way or a fight. This is not the OSS auth library. ## Verdict Auth-as-a-feature ships faster than auth-as-a-library. --- # Auth.js > Authentication for the web, framework-agnostic **Category:** Next.js **Pricing:** open-source **URL:** https://authjs.dev **GitHub:** https://github.com/nextauthjs/next-auth **Added:** 2026-08-18 --- ## The Hook Auth.js is authentication you own—with all the maintenance that implies. ## What's Good Provider support is comprehensive: OAuth, credentials, email magic links, and WebAuthn. The database adapter system works with your existing stack. The v5 rewrite improved Edge Runtime support and TypeScript types significantly. ## The Catch You are building the UI, handling edge cases, and managing the infrastructure. Provider quirks require debugging. ## Verdict Maximum control for teams willing to own their auth. --- # UploadThing > File uploads that do not make you cry **Category:** Next.js **Pricing:** freemium **URL:** https://uploadthing.com **GitHub:** https://github.com/pingdotgg/uploadthing **Added:** 2026-08-18 --- ## The Hook UploadThing turns file uploads into twenty lines of code. ## What's Good The API is elegant: define file routes with validation, use the React components, done. TypeScript inference flows through the entire stack. Built-in image optimization, middleware hooks for auth, and presigned uploads handle the common cases. ## The Catch You are depending on a third-party service for a core feature. Storage costs scale with usage, and egress gets expensive. ## Verdict Delightful until you need S3-level control. --- # Resend > Email API designed for developers **Category:** Next.js **Pricing:** freemium **URL:** https://resend.com **Added:** 2026-08-18 --- ## The Hook Resend made email feel like a modern API for the first time. ## What's Good Building emails with React components instead of table-based HTML is genuinely better. The SDK is minimal and typed. Delivery analytics work, and the testing flow catches issues early. ## The Catch Newer service means less track record. Deliverability depends on your domain reputation as much as theirs. Feature set is smaller than established players. ## Verdict Email finally feels like a modern API. --- # Neon > Serverless Postgres that branches **Category:** Next.js **Pricing:** freemium **URL:** https://neon.tech **Added:** 2026-08-18 --- ## The Hook Neon is Postgres that branches like git. ## What's Good Database branching works like git branching—create a copy-on-write branch for testing, preview deployments, or data experiments. Scale to zero saves money on development databases. The connection pooler handles serverless function connection limits. ## The Catch Cold starts affect latency for scale-to-zero databases. The branching model has a learning curve. Regional availability is limited. ## Verdict Postgres for the serverless era, with game-changing branching. --- # PlanetScale > Serverless MySQL with Vitess DNA **Category:** Next.js **Pricing:** freemium **URL:** https://planetscale.com **Added:** 2026-08-18 --- ## The Hook PlanetScale is YouTube-scale MySQL for the rest of us. ## What's Good The deploy request workflow makes schema changes safe—preview changes, review diff, deploy without downtime. Branching enables development workflows similar to git. Connection handling is serverless-friendly. ## The Catch No foreign key constraints enforced by the database—application-level enforcement required. The Vitess heritage means some MySQL features work differently. ## Verdict MySQL that scales like the big apps, with safer deploys. --- # Inngest > Background jobs and workflows with zero infrastructure **Category:** Next.js **Pricing:** freemium **URL:** https://inngest.com **GitHub:** https://github.com/inngest/inngest **Added:** 2026-08-19 --- ## The Hook Inngest makes serverless background jobs possible without running workers. ## What's Good Define functions in your codebase, deploy them to Vercel or any platform, and Inngest handles the queueing, retries, and execution. Step functions enable complex workflows. The local dev server shows exactly what is happening. TypeScript SDK is excellent. ## The Catch Another service dependency. Debugging distributed workflows requires understanding the execution model. Costs scale with function runs. ## Verdict The missing piece for serverless background processing. --- # Trigger.dev > Open-source background jobs for Next.js **Category:** Next.js **Pricing:** freemium **URL:** https://trigger.dev **GitHub:** https://github.com/triggerdotdev/trigger.dev **Added:** 2026-08-19 --- ## The Hook Trigger.dev is the open-source alternative to serverless job queues. ## What's Good Long-running tasks, scheduled jobs, and webhooks in one SDK. Self-hostable if you need it. The v3 rewrite brought a cleaner API and better dev experience. Integrations with common services work out of the box. ## The Catch v3 was a significant breaking change. The managed platform costs scale with compute. Self-hosting requires maintaining infrastructure. ## Verdict Open-source background jobs with a self-host escape hatch. --- # Convex > Reactive TypeScript backend with real-time sync **Category:** Next.js **Pricing:** freemium **URL:** https://www.convex.dev **Added:** 2026-08-19 --- ## The Hook Convex is the reactive TypeScript backend that syncs in real-time. ## What's Good Write backend logic in TypeScript, get real-time sync to the client automatically. No REST endpoints to define, no WebSocket plumbing to maintain. Queries are reactive—data changes, UI updates. The developer experience is genuinely different from traditional backends. Strong typing end-to-end. ## The Catch Not Postgres. Convex has its own data model and query language. Migration from existing databases requires rethinking your data layer. Chef (their AI feature) was deprecated. Vendor dependency is real. ## Verdict Reactive backend that feels like the future. Not Postgres. --- # Better Auth > TypeScript auth that lives in your app, not a rented IdP **Category:** Next.js **Pricing:** freemium **URL:** https://www.better-auth.com **GitHub:** https://github.com/better-auth/better-auth **Added:** 2026-09-03 --- ## The Hook Better Auth is the MIT library that replaced Auth.js for people who still want to own the database. ## What's Good Framework is MIT (LICENSE.md). Credentials, social, orgs, passkeys, MCP/agent auth, npx auth init. Hosted dashboard Starter is 0 USD/month (1 seat, 10,000 audit logs/month, 1-day retention). You keep the user table. Works next to Polar. Homepage names OpenAI, Databricks, and Strapi. ## The Catch The library is free; dashboards, audit, and SSO add-ons are the paid wedge. Starter audit retention is 1 day. This is not a hosted IdP — you still own the database and the incident. ## Verdict The MIT auth library to reach for before Clerk. Framework is free; the dashboard is the upsell. --- # Logto > MPL-2.0 identity platform with a 50k-MAU free Cloud **Category:** Next.js **Pricing:** freemium **URL:** https://logto.io **GitHub:** https://github.com/logto-io/logto **Added:** 2026-09-03 --- ## The Hook Logto is Auth0-shaped identity you can self-host, with a Cloud that is actually $0 to start. ## What's Good MPL-2.0. OIDC / OAuth 2.1 / SAML for SaaS: orgs, RBAC, SSO. Cloud Free is 0 USD/mo, no credit card, up to 50,000 MAU. OSS self-host is in the README. Pro from 24 USD/mo. ## The Catch Cloud Free is for starting out. Passkeys, hiding branding, orgs, MFA, and enterprise SSO live as Pro add-ons. MPL-2.0 is file-level copyleft, not MIT. ## Verdict The MPL-2.0 Auth0-shaped stack. Use it when Clerk MAUs are the product. --- # SuperTokens > Apache-2.0 session auth you can self-host with no MAU cap **Category:** Next.js **Pricing:** freemium **URL:** https://supertokens.com **GitHub:** https://github.com/supertokens/supertokens-core **Added:** 2026-09-03 --- ## The Hook SuperTokens is session-centric auth that does not meter your self-hosted MAUs. ## What's Good Apache-2.0 outside ee/. Frontend SDK + backend SDK + Java core. Passwordless, social, MFA, multi-tenancy. Pricing page: self-hosting the open-source features is free without any limits. ## The Catch Three processes, not a single library. ee/ is not Apache. Cloud is MAU-metered; MFA and account-linking have 100 USD/month minima on Cloud. Read the carve-out before you assume the whole repo is Apache-2.0. ## Verdict Apache-2.0 auth with no self-host MAU cap. Architecture is three pieces. --- # authentik > MIT-core self-hosted IdP; enterprise/ is not MIT **Category:** Next.js **Pricing:** open-source **URL:** https://goauthentik.io **GitHub:** https://github.com/goauthentik/authentik **Added:** 2026-09-03 --- ## The Hook authentik is the IdP you self-host when Okta is the lock-in and Keycloak is the other mountain. ## What's Good MIT outside enterprise/. OIDC, SAML, LDAP, RADIUS, proxy. OSS self-host is Free. Docker Compose or Kubernetes. Flows you can actually read. ## The Catch enterprise/ is not MIT — EE license, paid seats. No hosted hatch: goauthentik.io/pricing says they do not currently provide a hosted version. You run Postgres and the stack. ## Verdict MIT-core IdP. Self-host is the free path; there is no authentik Cloud. --- # Hanko > AGPLv3 passkeys-first auth with a 10k-MAU free Cloud **Category:** Next.js **Pricing:** freemium **URL:** https://www.hanko.io **GitHub:** https://github.com/teamhanko/hanko **Added:** 2026-09-03 --- ## The Hook Hanko is passkeys first, passwords optional, and the Cloud actually starts at $0. ## What's Good AGPLv3 core. Passkeys, passcodes, optional passwords, social, SAML. Cloud Starter is Free: 10,000 MAU, no credit card, at hanko.io/pricing. Self-host the backend. Elements and frontend SDKs exist. ## The Catch Split licenses: backend AGPLv3, Elements and frontend SDKs MIT. Cloud Starter Free 10k MAU is the wall — after that you pay. Cloud Console is closed source. AGPL on the API is network copyleft. ## Verdict Passkeys-first auth. 10k MAU is free; the license split is the other catch. --- # ZITADEL > AGPLv3 identity Cloud at US$0/Month — 100 DAU, not Logto 50k MAU **Category:** Next.js **Pricing:** freemium **URL:** https://zitadel.com **GitHub:** https://github.com/zitadel/zitadel **Added:** 2026-09-03 --- ## The Hook ZITADEL is the AGPL IdP whose Cloud is actually US$0 — until the 100th daily user. ## What's Good AGPLv3. Cloud Free is US$0/Month: 100 Daily Active Users, unlimited stored users, 3 identity providers, unlimited organizations. Data residency EU, US, Switzerland, Australia. Self-host the same stack. Not Logto (50k MAU Cloud), not Hanko (10k MAU passkeys), not authentik (MIT-core, no Cloud). ## The Catch Loud: the wall is 100 DAU, not Logto 50k MAU. AGPLv3 is network copyleft — Cloud or self-host, the license is loud. Pro is US$100/Month for 25k DAU. Distinct from Hanko passkeys-first Cloud and authentik no-hosted-hatch. ## Verdict AGPL identity with a real $0 Cloud. 100 DAU is the meter; self-host if that is a joke. --- # Kinde > Hosted auth plus billing, $0 forever, 10,500 MAU, no OSS **Category:** Next.js **Pricing:** freemium **URL:** https://kinde.com **Added:** 2026-09-03 --- ## The Hook Kinde is Clerk-shaped auth that also meters subscriptions — and Free is actually $0 forever. ## What's Good Hosted, no OSS product. Free is $0 forever, no credit card, 10,500 MAU. Auth plus subscriptions/billing on Free. Billing take-rate 0.7%. Users charged more than USD $4/month via Kinde do not count as MAU. MFA, social, orgs, custom domain on Free. ## The Catch Not self-host. Kinde branding stays on Free. SMS is 10 included. Pro is $25/mo to uncap MAU and drop branding. Distinct from Polar (Apache MoR, not an IdP) and Clerk (auth UI, no 0.7% billing take-rate on Free). ## Verdict Hosted auth+billing with a real $0 forever plan. No repo to fork; branding until Pro. --- # Authelia > Apache-2.0 reverse-proxy SSO, not a full IdP like authentik **Category:** Next.js **Pricing:** open-source **URL:** https://www.authelia.com **GitHub:** https://github.com/authelia/authelia **Added:** 2026-09-03 --- ## The Hook Authelia is the SSO portal you put behind Traefik when authentik is the other mountain. ## What's Good Apache-2.0. OSS self-host, no product pricing page. OpenID Connect 1.0 Provider (OpenID Certified). Passkeys, TOTP, WebAuthn, policies. Companion for Traefik, Caddy, nginx. Go + React, tiny container. ## The Catch You run it behind Traefik/Caddy/nginx. Thin proxy SSO, not a full IdP like authentik (flows, LDAP/RADIUS/SAML, enterprise/). No hosted Authelia Cloud. Ops, the user store, and the proxy stay yours. ## Verdict Apache-2.0 proxy SSO. Free as in you operate the reverse proxy. --- # Keycloak > Apache-2.0 CNCF-incubation Java IAM: OIDC, OAuth, SAML, LDAP **Category:** Next.js **Pricing:** open-source **URL:** https://www.keycloak.org **GitHub:** https://github.com/keycloak/keycloak **Added:** 2026-09-03 --- ## The Hook Keycloak is the CNCF IdP the JVM shops already run — not authentik with a different coat. ## What's Good Apache-2.0. CNCF incubation. OpenID Connect, OAuth 2.0, SAML 2.0, LDAP user federation, identity brokering. Admin console, realms, clients. Self-host is the product. ## The Catch CNCF Java IAM, OIDC+OAuth+SAML+LDAP. Distinct from authentik (MIT-core Python, no hosted hatch, enterprise/ carve-out). You run a JVM and a database. There is no $0 Keycloak Cloud on the project site. ## Verdict Apache-2.0 CNCF IdP. Heavy, complete, self-host. Not authentik. --- # Dex > Apache-2.0 federated OIDC IdP that owns no user state **Category:** Next.js **Pricing:** open-source **URL:** https://dexidp.io **GitHub:** https://github.com/dexidp/dex **Added:** 2026-09-03 --- ## The Hook Dex is the OIDC adapter in front of GitHub, LDAP, or SAML — it does not keep your users. ## What's Good Apache-2.0. Federated OIDC / OAuth 2.0 provider with pluggable connectors (GitHub, Google, LDAP, SAML, OIDC). Apps speak OIDC to Dex once. Go binary, Helm chart, CNCF sandbox. ## The Catch Protocol adapter, not a user directory and not a proxy gate. Distinct from Authelia (reverse-proxy SSO) and from authentik/ZITADEL/Keycloak (they store users). Dex owns no user state. You still need an upstream IdP. ## Verdict Apache-2.0 OIDC facade. Connectors in, standards out, no user table. --- # Cerbos > Apache-2.0 authZ PDP: YAML policies, run anywhere **Category:** Next.js **Pricing:** open-source **URL:** https://cerbos.dev **GitHub:** https://github.com/cerbos/cerbos **Added:** 2026-09-04 --- ## The Hook Cerbos is the YAML policy decision point when your IdP should not also be your authorization brain. ## What's Good Apache-2.0. Free forever Run anywhere — verified on cerbos.dev/pricing. YAML policies, PDP you embed or sidecars. Decouples authZ from whatever issues the tokens. ## The Catch authZ YAML PDP — not authentik/Keycloak/Better Auth authN/IdP. Do not reprint Hub prices; self-host PDP is the free path. Distinct from authentik, Keycloak, and Better Auth. ## Verdict Apache-2.0 authorization PDP. Free to run; not an IdP. --- # OpenFGA > Apache-2.0 Zanzibar-style ReBAC with a Check API **Category:** Next.js **Pricing:** open-source **URL:** https://openfga.dev **GitHub:** https://github.com/openfga/openfga **Added:** 2026-09-04 --- ## The Hook OpenFGA is Google Zanzibar-shaped authorization when YAML ABAC is the wrong model. ## What's Good Apache-2.0. Relationship-based access control, Check API, modeling language. Self-host the authorization service. CNCF-flavored ReBAC without inventing your own graph. ## The Catch Zanzibar-style ReBAC + Check API — not Cerbos YAML ABAC. Vs SpiceDB: CNCF/Okta OpenFGA modeling + Check API lane; SpiceDB is AuthZed production Zanzibar + ZedTokens/Cloud. Distinct from Cerbos. ## Verdict Apache-2.0 ReBAC. Self-host free. Not Cerbos YAML. --- # Topaz > Apache-2.0 authZ: OPA Rego + embedded Zanzibar in one container **Category:** Next.js **Pricing:** open-source **URL:** https://www.topaz.sh **GitHub:** https://github.com/aserto-dev/topaz **Added:** 2026-09-04 --- ## The Hook Topaz is authorization in one container when you refuse to pick only ABAC or only ReBAC. ## What's Good Apache-2.0. OPA Rego policies plus an embedded Zanzibar-style directory in a single binary/container. ABAC and ReBAC without stitching Cerbos to OpenFGA yourself. Self-host is the free path. ## The Catch One container = OPA Rego + embedded Zanzibar directory (ABAC+ReBAC) — not Cerbos YAML-only PDP and not OpenFGA ReBAC-only Check API. Distinct from Cerbos and OpenFGA. ## Verdict Apache-2.0 combined authZ. Self-host free; not a single-model PDP. --- # Unleash > AGPLv3 feature flags; OSS caps 1 project / 2 environments **Category:** Next.js **Pricing:** open-source **URL:** https://www.getunleash.io/open-source **GitHub:** https://github.com/Unleash/unleash **Added:** 2026-09-04 --- ## The Hook Unleash is AGPLv3 feature flags you self-host when PostHog flags are bolted onto analytics. ## What's Good AGPLv3. Dedicated feature-flag server. Self-host is the free path. SDKs across stacks. Open-source page is the pitch — not Cloud trial. ## The Catch Loud: OSS caps Projects 1 / Environments 2. Cloud = Enterprise trial FAIL — pitch self-host only. Vs PostHog dedicated-flags-inside-analytics. Distinct from PostHog. ## Verdict AGPLv3 flags. Self-host with OSS caps; Cloud trial is not free forever. --- # Flagsmith > BSD-3-Clause flags; Cloud Free 50k req/mo, self-host uncapped **Category:** Next.js **Pricing:** freemium **URL:** https://www.flagsmith.com **GitHub:** https://github.com/Flagsmith/flagsmith **Added:** 2026-09-04 --- ## The Hook Flagsmith is BSD-3 feature flags with a real Cloud Free and uncapped self-host. ## What's Good BSD-3-Clause. Cloud Free: 50k requests/mo, 1 seat, 1 project. Self-host uncapped. Flags + remote config without PostHog warehouse story. ## The Catch Cloud Free is capped (50k/1 seat/1 project); self-host uncapped. Vs PostHog analytics-native flags and Unleash AGPL OSS caps. Distinct from PostHog and Unleash. ## Verdict BSD-3 flags. Cloud Free is real but capped; self-host for scale. --- # GrowthBook > MIT-core flags+experiments; Cloud Starter Free, self-host unlimited users **Category:** Next.js **Pricing:** freemium **URL:** https://growthbook.io **GitHub:** https://github.com/growthbook/growthbook **Added:** 2026-09-04 --- ## The Hook GrowthBook is MIT-core flags and A/B when PostHog owns the warehouse-native analytics seat. ## What's Good MIT core. Cloud Starter Free — no CC — unlimited flags/experiments (3 users / 1 project). Self-Hosted OSS unlimited users. Experiment stats engine included. Do not reprint Pro seat prices. ## The Catch Warehouse-native experimentation competitor to PostHog — different pitch (flags+experiments first). Cloud Starter Free has user/project caps; self-host OSS uncapped users. Distinct from PostHog. ## Verdict MIT-core flags+experiments. Cloud Starter Free is real; ignore Pro seats. --- # WorkOS > Proprietary AuthKit: first 1M MAUs Free — LOUD not OSS **Category:** Next.js **Pricing:** freemium **URL:** https://workos.com/pricing **Added:** 2026-09-04 --- ## The Hook WorkOS is enterprise-ready AuthKit when Clerk is consumer auth and you need SSO/Directory Sync later. ## What's Good AuthKit first 1M MAUs Free; staging free. SSO, Directory Sync, Audit Log paths when you outgrow DIY. Polished Next.js integration. Do not reprint Enterprise monthly prices. ## The Catch LOUD: proprietary — not OSS. Vs Clerk/Logto/Kinde/Hanko/Better Auth. Free MAU band is the pitch; paid enterprise features exist — do not invent numbers. Distinct from Clerk, Logto, Kinde, Hanko, and Better Auth. ## Verdict Proprietary AuthKit. First 1M MAUs Free; not open source. --- # Descope > Proprietary auth: Free Forever 7500 MAUs / 10 tenants / 3 SSO **Category:** Next.js **Pricing:** freemium **URL:** https://www.descope.com/pricing **Added:** 2026-09-04 --- ## The Hook Descope is drag-drop auth flows when you want Free Forever MAUs without standing up Keycloak. ## What's Good Free Forever: 7500 MAUs / 10 tenants / 3 SSO. Visual flows, passkeys, SSO on the free band. Fast to ship customer auth. Skills already live; gem was absent. ## The Catch LOUD: proprietary — not OSS. Free Forever caps are real; above them is paid. Vs Clerk/Logto/Kinde/WorkOS/Hanko/Better Auth. Distinct from Clerk, Logto, Kinde, WorkOS, Hanko, and Better Auth. ## Verdict Proprietary auth. Free Forever 7.5k MAUs — not open source. --- # Casdoor > Apache-2.0 UI-first IdP; Hosted Free 25k MAU, self-host primary **Category:** Next.js **Pricing:** freemium **URL:** https://casdoor.com **GitHub:** https://github.com/casdoor/casdoor **Added:** 2026-09-04 --- ## The Hook Casdoor is Apache-2.0 identity with a web UI when Keycloak feels like an ops novel. ## What's Good Apache-2.0. Self-host is primary. Hosted Free: 25k MAU. OAuth/OIDC/SAML, UI for apps and users. Use casdoor.com (casdoor.org 500s). ## The Catch UI-first IdP — vs Keycloak/authentik/Kanidm/Logto/Hanko/Zitadel. Hosted Free 25k MAU; self-host uncapped. Distinct from Keycloak, authentik, Kanidm, Logto, Hanko, and Zitadel. ## Verdict Apache-2.0 IdP. Self-host primary; Hosted Free 25k MAU. --- # SimpleWebAuthn > MIT passkeys ceremony library — not an IdP **Category:** Next.js **Pricing:** open-source **URL:** https://simplewebauthn.dev **GitHub:** https://github.com/MasterKale/SimpleWebAuthn **Added:** 2026-09-04 --- ## The Hook SimpleWebAuthn is the WebAuthn ceremony kit when you already have an IdP and only need passkeys wired correctly. ## What's Good MIT. Server and browser helpers for registration/authentication ceremonies. You own sessions and users. No hosted IdP tax for the library itself. ## The Catch Passkeys ceremony lib — not Hanko/Better Auth/Auth.js IdP or auth framework. You still store credentials and issue sessions. Distinct from Hanko, Better Auth, and Auth.js. ## Verdict MIT WebAuthn helpers. Library only; bring your own IdP. --- # Infisical > MIT secrets; Cloud Free forever 5 identities; ee/ needs LICENSE_KEY **Category:** Next.js **Pricing:** freemium **URL:** https://infisical.com/pricing **GitHub:** https://github.com/Infisical/infisical **Added:** 2026-09-04 --- ## The Hook Infisical is secrets with a PostHog-shaped free path when Vault is the wrong shape. ## What's Good MIT core. Cloud Free $0 forever: 5 identities, unlimited projects, 3 envs, 10 secret syncs. MIT self-host uncapped at $0. Skills already on the catalog; gem was missing. ## The Catch ee/ needs LICENSE_KEY even self-hosted (SSO+, versioning, rotation, dynamic secrets) — authentik enterprise pattern. Stretch: secrets remit. Distinct from authentik (IdP) and from Vault-scale ops. ## Verdict MIT secrets with a real Free forever Cloud hatch. ee/ is the paid carve-out. --- # OpenBao > MPL-2.0 Vault-compatible secrets engine after the Vault fork **Category:** Next.js **Pricing:** open-source **URL:** https://openbao.org **GitHub:** https://github.com/openbao/openbao **Added:** 2026-09-04 --- ## The Hook OpenBao is the Vault-shaped secrets engine you run when the HashiCorp license line is the problem. ## What's Good MPL-2.0. Linux Foundation Vault-compatible fork. KV, transit, auth methods, policies — the engine shape teams already know. Self-host is the free path. ## The Catch Vault-engine / IdP-secrets lane, not Infisical app-secrets DX. Infisical already shipped — this is the engine/compat fork, not the developer-secrets UI. Do not invent Cloud free tiers. ## Verdict MPL Vault-compat engine. Fork for license, not Infisical DX. --- # Ory Open Source > Apache-2.0 modular Go IAM: Kratos, Hydra, Keto, Oathkeeper **Category:** Next.js **Pricing:** open-source **URL:** https://www.ory.com/open-source **GitHub:** https://github.com/ory/kratos **Added:** 2026-09-04 --- ## The Hook Ory Open Source is the modular Go IAM stack you compose instead of adopting one Java realm. ## What's Good Apache-2.0. Self-host Kratos (identity), Hydra (OAuth2/OIDC), Keto (Zanzibar authz), Oathkeeper (IAP). Pick the pieces you need. Same APIs as OEL/Network. ## The Catch Vs Keycloak: modular Go you compose, not one Java IAM. LOUD: Ory own compare table still labels Ory Open Source Evaluation and prototyping; FAQ steers prototypes to OSS and scale to OEL/Network — print that. Ory Cloud/Network CUT as the free path. Not a Dex/Keycloak/Casdoor twin if the Catch stays modular-compose. ## Verdict Apache-2.0 modular Go IAM. Compose pieces; their table still says prototyping. --- # Pocket ID > BSD-2-Clause passkey-only OIDC IdP; self-host only **Category:** Next.js **Pricing:** open-source **URL:** https://pocket-id.org **GitHub:** https://github.com/pocket-id/pocket-id **Added:** 2026-09-04 --- ## The Hook Pocket ID is the passkey-only OIDC IdP when Hanko still offers passwords and Authelia is proxy SSO. ## What's Good BSD-2-Clause. Self-host only — no paid Cloud trap. OIDC Certified. Passkeys only: no password store to mismanage. Homelab and MCP-scoped auth lane. ## The Catch Vs Hanko/Authelia/Keycloak/Ory: passkey-ONLY OIDC IdP (no passwords). Homelab/MCP-scoped, not enterprise IAM. Free path is self-host only. ## Verdict BSD-2-Clause passkey-only OIDC. Self-host; not enterprise IAM. --- # Tinyauth > AGPL-3.0 one-binary proxy auth + Certified OIDC **Category:** Next.js **Pricing:** open-source **URL:** https://tinyauth.app **GitHub:** https://github.com/steveiliop56/tinyauth **Added:** 2026-09-04 --- ## The Hook Tinyauth is the single statically linked binary you put in front of Traefik when Authelia is the heavier portal. ## What's Good AGPL-3.0. Self-host forever; single statically linked binary, env-var config. OIDC Certified (Basic OP). Works with Traefik/Nginx/Caddy plus LDAP/OAuth. ## The Catch Vs Authelia/Pocket ID: tiny proxy-fronted auth + Certified OIDC in one binary — not Authelia reverse-proxy SSO alone, not passkeys-only Pocket ID. Free path is self-host. ## Verdict AGPL one-binary proxy auth + Certified OIDC. Self-host forever. --- # SpiceDB > Apache-2.0 Google Zanzibar permissions DB; AuthZed Cloud optional **Category:** Next.js **Pricing:** freemium **URL:** https://authzed.com/spicedb/ **GitHub:** https://github.com/authzed/spicedb **Added:** 2026-09-04 --- ## The Hook SpiceDB is the AuthZed Zanzibar permissions database when OpenFGA is the CNCF twin and Cerbos is YAML ABAC. ## What's Good Apache-2.0. Relationship-based authZ graph faithful to Google Zanzibar. Schema language, ZedTokens for consistency, Postgres/Cockroach/Spanner/MySQL stores. Self-host free; AuthZed Cloud is usage-based when you do not want to operate it. ## The Catch Vs OpenFGA/Cerbos/Topaz: AuthZed production Zanzibar + ZedTokens/Cloud — not OpenFGA CNCF Check API twin, not Cerbos YAML PDP, not Topaz Rego+ReBAC bundle. Free path is self-host; Cloud is paid usage. ## Verdict Apache-2.0 Zanzibar permissions DB. Self-host free; AuthZed Cloud optional. --- # Payload > MIT Next.js-native headless CMS; Cloud new deploys paused **Category:** Next.js **Pricing:** open-source **URL:** https://payloadcms.com **GitHub:** https://github.com/payloadcms/payload **Added:** 2026-09-04 --- ## The Hook Payload is the code-first Next.js CMS when Strapi is the Node admin cousin and Sanity is the hosted Content Lake. ## What's Good MIT. Collections/globals in TypeScript, Auth, uploads, Local API, runs inside Next.js. Self-host forever on any Node host. Active core post-Figma acquisition with public commitment to stay OSS. ## The Catch LOUD: Payload Cloud new project deploys are paused after Figma — existing Cloud keeps running; new teams should self-host (Vercel/Cloudflare templates). Not Sanity Content Lake SaaS. Enterprise features/support are commercial conversations. ## Verdict MIT Next.js CMS. Self-host is the free path; Cloud new-signups paused. --- # FusionAuth > Source-available CIAM; Community self-host free unlimited users **Category:** Next.js **Pricing:** freemium **URL:** https://fusionauth.io **Added:** 2026-09-04 --- ## The Hook FusionAuth is the downloadable auth server when Keycloak is the OSS Java IAM and Auth0 is the per-MAU SaaS bill. ## What's Good Community Edition: free forever self-host, unlimited users/MAU — OIDC/OAuth/SAML, MFA basics, themes, APIs, Lambdas, multi-tenancy. Flat paid tiers (not per-MAU) for advanced MFA, breached-password, SCIM, theming extras, support. Cloud hosting optional. ## The Catch LOUD: NOT open source — source-available proprietary CE. Vs Keycloak/ZITADEL/Ory: easier admin UX + flat pricing, but you cannot fork freely like Apache/AGPL IdPs. Premium features and FusionAuth Cloud are paid. Ops burden is yours on CE. ## Verdict Free unlimited self-host CIAM. Not OSS; paid for premium/Cloud. --- # Stytch > Auth SaaS with no feature gates; free to 10k MAU + 5 SSO/SCIM **Category:** Next.js **Pricing:** freemium **URL:** https://stytch.com **Added:** 2026-09-04 --- ## The Hook Stytch is hosted auth without the Auth0 feature cliff when Clerk is consumer-polished and WorkOS is enterprise-SSO-first. ## What's Good Pay-as-you-go from $0. Free includes 10k MAU (users+AI agents), unlimited orgs, 5 SSO/SCIM, 1k M2M tokens — MFA/SSO/RBAC/SCIM not gated. B2B multi-tenancy, admin portal components, optional fraud tooling. ## The Catch Closed SaaS — not Better Auth/Ory self-host. LOUD: SSO/SCIM beyond 5 is $125/connection/mo; brand removal $99/mo; fingerprints $0.005 after 10k. MAU overage metered (no hard cliff). HIPAA/SLA need Enterprise. ## Verdict Hosted auth, features unlocked. Free to 10k MAU/5 SSO; usage meters after. --- # ◆ Python Modern Python development: fast runtimes, type-safe frameworks, and tools that respect your time. *13 gems in this category* --- # FastAPI > Modern Python APIs at startup speed **Category:** Python **Pricing:** open-source **URL:** https://fastapi.tiangolo.com **GitHub:** https://github.com/tiangolo/fastapi **Added:** 2026-08-18 --- ## The Hook FastAPI is the Python API framework you will actually enjoy using. ## What's Good Pydantic integration means request/response validation is declarative and automatic. OpenAPI docs generate from type hints. Async support is first-class. The dependency injection system is powerful without being magical. ## The Catch The ecosystem assumes you will integrate pieces yourself—no batteries-included ORM or auth. Async programming has its own learning curve. ## Verdict The Python API framework you will actually enjoy. --- # Pydantic > Data validation using Python type hints **Category:** Python **Pricing:** open-source **URL:** https://docs.pydantic.dev **GitHub:** https://github.com/pydantic/pydantic **Added:** 2026-08-18 --- ## The Hook Pydantic is Python data validation solved. Just use it. ## What's Good V2 Rust core made it dramatically faster. Settings management handles environment variables elegantly. JSON Schema generation enables interop with other tools. The ecosystem integration (FastAPI, LangChain, SQLModel) means Pydantic models flow through your entire stack. ## The Catch The v1 to v2 migration was painful—significant API changes broke existing code. Some advanced validation patterns require understanding the internals. ## Verdict Python data validation solved. Just use it. --- # uv > Python packaging at Rust speed **Category:** Python **Pricing:** open-source **URL:** https://github.com/astral-sh/uv **GitHub:** https://github.com/astral-sh/uv **Added:** 2026-08-18 --- ## The Hook uv makes Python packaging not terrible for the first time. ## What's Good Installation is 10-100x faster than pip. The resolver handles complex dependency trees without the random failures. Virtual environment management is built-in. Compatible with requirements.txt and pyproject.toml. ## The Catch Relatively new, so edge cases exist. Some older packages with unusual build systems may have issues. ## Verdict Finally, Python packaging that does not waste your time. --- # Ruff > Python linter faster than your test suite **Category:** Python **Pricing:** open-source **URL:** https://docs.astral.sh/ruff/ **GitHub:** https://github.com/astral-sh/ruff **Added:** 2026-08-18 --- ## The Hook Ruff replaces your entire Python lint stack and runs faster than any piece of it. ## What's Good The rule coverage is comprehensive: 800+ rules from popular linters. Auto-fix handles most issues. Configuration is straightforward. The formatter (ruff format) now handles code style too. ## The Catch Not every rule from every tool is implemented. Some teams have strong opinions about specific linter behaviors. ## Verdict One fast tool to replace your entire lint toolchain. --- # Polars > DataFrames that respect your CPU **Category:** Python **Pricing:** open-source **URL:** https://pola.rs **GitHub:** https://github.com/pola-rs/polars **Added:** 2026-08-18 --- ## The Hook Polars is what pandas would be if designed today. ## What's Good The API is expressive: method chaining, lazy queries, and automatic parallelization. Memory efficiency is dramatically better than pandas for large datasets. The query optimizer makes naive code fast. ## The Catch Different API than pandas—existing codebases need rewriting. The ecosystem is smaller; some pandas integrations do not exist. ## Verdict What pandas would be if designed today. --- # Typer > CLI apps with type hints and zero effort **Category:** Python **Pricing:** open-source **URL:** https://typer.tiangolo.com **GitHub:** https://github.com/tiangolo/typer **Added:** 2026-08-18 --- ## The Hook Typer turns type hints into a CLI without you doing anything. ## What's Good Define a function with type hints; Typer generates the CLI with help text, validation, and shell completion. Rich integration provides beautiful output. Nested commands work intuitively. ## The Catch Complex CLI patterns sometimes fight the abstraction. Very dynamic interfaces may need lower-level Click access. ## Verdict CLI tools in minutes instead of hours. --- # Rich > Beautiful terminal output without the pain **Category:** Python **Pricing:** open-source **URL:** https://rich.readthedocs.io **GitHub:** https://github.com/Textualize/rich **Added:** 2026-08-18 --- ## The Hook Rich makes terminals look like they belong in 2026. ## What's Good The Console class handles everything: print replacement, styling, layouts. Progress bars work for downloads, processing, and iteration. Tracebacks are gorgeous and more readable. ## The Catch Terminal compatibility varies—some environments render poorly. The feature set is large; learning the full API takes time. ## Verdict The easiest way to make terminals beautiful. --- # HTTPX > Async HTTP client for modern Python **Category:** Python **Pricing:** open-source **URL:** https://www.python-httpx.org **GitHub:** https://github.com/encode/httpx **Added:** 2026-08-18 --- ## The Hook HTTPX is requests for the async era. ## What's Good Sync and async clients share the same interface. HTTP/2 support is built-in. Connection pooling and timeouts work sensibly. The test client enables easy mocking. ## The Catch Another dependency when requests works for simple cases. Some requests plugins do not have HTTPX equivalents. ## Verdict The requests replacement for the async era. --- # SQLModel > SQL + Pydantic in one elegant package **Category:** Python **Pricing:** open-source **URL:** https://sqlmodel.tiangolo.com **GitHub:** https://github.com/tiangolo/sqlmodel **Added:** 2026-08-18 --- ## The Hook SQLModel is one model for both your database and your API. ## What's Good The inheritance from both SQLAlchemy and Pydantic means models work everywhere. Async support integrates cleanly with FastAPI. The relationship handling is intuitive for common patterns. ## The Catch Relatively young library, so edge cases exist. Complex SQLAlchemy patterns may need direct SQLAlchemy access. ## Verdict One model for database and API—elegant when it fits. --- # Litestar > The FastAPI alternative with batteries **Category:** Python **Pricing:** open-source **URL:** https://litestar.dev **GitHub:** https://github.com/litestar-org/litestar **Added:** 2026-08-18 --- ## The Hook Litestar is FastAPI with more opinions and more built-in. ## What's Good Built-in support for SQLAlchemy, Redis, and common patterns reduces integration work. The dependency injection is more powerful than FastAPI. OpenAPI generation works similarly. ## The Catch Smaller community means fewer tutorials and Stack Overflow answers. Some FastAPI extensions need equivalents. ## Verdict More opinionated FastAPI alternative worth considering. --- # Textual > Build TUIs like web apps **Category:** Python **Pricing:** open-source **URL:** https://textual.textualize.io **GitHub:** https://github.com/Textualize/textual **Added:** 2026-08-19 --- ## The Hook Textual brings CSS layouts and reactive widgets to the terminal. ## What's Good The component model feels familiar to web developers. CSS-style layouts work in the terminal. Async-first design handles I/O without blocking. The widget library covers common patterns. ## The Catch Terminal constraints limit what is possible. Learning curve for the reactive model. Not every terminal renders correctly. ## Verdict TUI development that feels like modern web dev. --- # ty > MIT Astral Python type checker (beta) in Rust **Category:** Python **Pricing:** open-source **URL:** https://docs.astral.sh/ty/ **GitHub:** https://github.com/astral-sh/ty **Added:** 2026-09-04 --- ## The Hook ty is Astral Rust type checker when mypy or Pyright is the baseline you are measuring against. ## What's Good MIT. Extremely fast Python type checker and language server from the uv/Ruff team. uvx ty check to try. Incremental analysis aimed at IDE speed. ## The Catch Beta/maturing vs mypy/pyright — say so. Do not invent pass rates or claim it already replaces either tool on every codebase. Distinct from Ruff (lint/format) even though same vendor. ## Verdict MIT Astral type checker. Fast and young — measure it; do not invent scores. --- # Instructor > MIT Pydantic structured outputs for any LLM; free library **Category:** Python **Pricing:** open-source **URL:** https://python.useinstructor.com **GitHub:** https://github.com/instructor-ai/instructor **Added:** 2026-09-04 --- ## The Hook Instructor is validated structured extraction when raw JSON mode and hand-rolled retries are the weekend you lose. ## What's Good MIT. Pydantic models → typed LLM outputs with auto-retry on validation failure. 15+ providers (OpenAI, Anthropic, Gemini, Ollama, …). Streaming/partials. Ports in TS/Go/Ruby/etc. Millions of monthly downloads. ## The Catch Library not an agent framework (pair with your orchestrator). Provider token costs remain. Vs native structured-output APIs: Instructor shines on validation/retry DX across providers, not on replacing tool-calling agents. ## Verdict MIT structured LLM outputs. Free forever; you pay model APIs. --- # ◇ Shopify Building commerce: headless storefronts, apps, themes, and everything merchants need to sell. *9 gems in this category* --- # Hydrogen > Shopify React framework for headless commerce **Category:** Shopify **Pricing:** open-source **URL:** https://hydrogen.shopify.dev **GitHub:** https://github.com/Shopify/hydrogen **Added:** 2026-08-18 --- ## The Hook Hydrogen is the headless path Shopify actually wants you to take. ## What's Good First-party Storefront API components handle cart, checkout, and product data. Oxygen hosting is tightly integrated. The Shopify CLI scaffolds projects with working commerce features. ## The Catch You are locked into Shopify architectural choices. Customization beyond their patterns requires deeper knowledge. Oxygen hosting limits deployment options. ## Verdict The official path for Shopify headless—for better and worse. --- # Remix > Full stack React that Shopify acquired **Category:** Shopify **Pricing:** open-source **URL:** https://remix.run **GitHub:** https://github.com/remix-run/remix **Added:** 2026-08-18 --- ## The Hook Remix brought web fundamentals back to React development. ## What's Good The loader/action pattern simplifies data flow compared to useEffect sprawl. Nested routing with parallel data loading improves performance. Error boundaries work at route level. ## The Catch The mental model shift from SPA patterns takes adjustment. Some React patterns do not translate directly. The ecosystem is smaller than Next.js. ## Verdict Web fundamentals meet React, now with Shopify backing. --- # Sanity > Headless CMS that developers actually like **Category:** Shopify **Pricing:** freemium **URL:** https://sanity.io **Added:** 2026-08-18 --- ## The Hook Sanity is structured content done right. ## What's Good The schema-as-code model means content types live in your repo. GROQ query language is expressive. Real-time collaboration works. The Studio is customizable and deployable. ## The Catch Another service to manage and pay for. The schema learning curve is real. Heavy customization requires understanding Sanity Studio internals. ## Verdict Structured content done right—worth the learning curve. --- # Shopify CLI > The development tool every Shopify dev needs **Category:** Shopify **Pricing:** open-source **URL:** https://shopify.dev/docs/api/shopify-cli **GitHub:** https://github.com/Shopify/cli **Added:** 2026-08-18 --- ## The Hook Shopify CLI is not optional—just install it. ## What's Good Theme development with hot reload changed the game. App development handles the OAuth tunnel. Hydrogen scaffolding works out of the box. Environment management simplifies working with multiple stores. ## The Catch Occasional bugs with authentication flows. Resource usage can be heavy. Some advanced workflows still require manual setup. ## Verdict Not optional—just install it and move on. --- # Polaris > Shopify design system for embedded apps **Category:** Shopify **Pricing:** open-source **URL:** https://polaris.shopify.com **GitHub:** https://github.com/Shopify/polaris **Added:** 2026-08-18 --- ## The Hook Polaris makes embedded apps look native to Shopify admin. ## What's Good Components cover common admin patterns: data tables, forms, modals, and navigation. The design tokens enable consistent theming. App Bridge integration handles embedded app requirements. ## The Catch Looks out of place outside Shopify admin. Breaking changes between major versions require updates. ## Verdict Mandatory for embedded apps, pointless elsewhere. --- # Checkout UI Extensions > Customize Shopify checkout safely **Category:** Shopify **Pricing:** open-source **URL:** https://shopify.dev/docs/api/checkout-ui-extensions **Added:** 2026-08-18 --- ## The Hook Checkout Extensions are the only supported way to customize Shopify checkout. ## What's Good Extension points cover key checkout moments: cart, shipping, payment, and order status. The component library ensures visual consistency. The API handles state without direct DOM manipulation. ## The Catch The sandboxed environment limits what is possible. Some desired customizations are not supported by extension points. ## Verdict The only way to customize checkout—work within the limits. --- # GraphQL Code Generator > Type-safe Shopify API without manual typing **Category:** Shopify **Pricing:** open-source **URL:** https://the-guild.dev/graphql/codegen **GitHub:** https://github.com/dotansimha/graphql-code-generator **Added:** 2026-08-18 --- ## The Hook GraphQL Codegen writes the TypeScript so you do not have to. ## What's Good The preset ecosystem handles common patterns: typed document nodes, React hooks, or pure types. Schema fetching from Shopify endpoints works. Watch mode regenerates on changes. ## The Catch Configuration complexity for advanced setups. Build step adds friction to the workflow. Large schemas generate large type files. ## Verdict Essential infrastructure for typed Shopify GraphQL. --- # Shopify App JS > Official Node.js library for Shopify apps **Category:** Shopify **Pricing:** open-source **URL:** https://shopify.dev/docs/apps/tools/shopify-app-js **GitHub:** https://github.com/Shopify/shopify-app-js **Added:** 2026-08-19 --- ## The Hook Shopify App JS handles OAuth so you do not have to think about it. ## What's Good Session storage, webhook handling, and API client in one package. Works with Express, Koa, or any Node framework. The Remix adapter makes Hydrogen app development smoother. ## The Catch Node.js only. Some edge cases require diving into implementation details. Version updates can break existing apps. ## Verdict The foundation every Node Shopify app needs. --- # Saleor > BSD-3 GraphQL headless commerce; self-host free, Cloud from $1599/mo **Category:** Shopify **Pricing:** freemium **URL:** https://saleor.io **GitHub:** https://github.com/saleor/saleor **Added:** 2026-09-04 --- ## The Hook Saleor is Python/GraphQL headless commerce when Shopify Hydrogen is the hosted storefront stack and Medusa is the Node twin. ## What's Good BSD-3-Clause — one OSS core, no feature fragmentation. GraphQL-only API, multichannel, webhooks, composable storefronts. Free Cloud sandboxes for DIY. Self-host forever on your infra. ## The Catch Vs Shopify/Hydrogen: you own ops or pay Cloud. LOUD: production Cloud Select from about $1599/mo (to ~$200k GMV) / Volume ~$3999/mo — sandboxes are not for live selling. GraphQL-only + Django stack is a team bet. Not a Shopify App Store ecosystem. ## Verdict BSD-3 headless commerce. Self-host free; Cloud production starts ~$1599/mo. --- # ⬡ Databases Data infrastructure that scales: SQL, NoSQL, vector stores, and the tools that make them manageable. *21 gems in this category* --- # Supabase > Open-source Firebase alternative built on Postgres **Category:** Databases **Pricing:** freemium **URL:** https://supabase.com **GitHub:** https://github.com/supabase/supabase **Added:** 2026-08-19 --- ## The Hook Supabase is Postgres with auth, realtime, and storage bolted on. ## What's Good Real Postgres underneath means you are not locked in. Auth, storage, edge functions, and realtime subscriptions in one platform. The dashboard is genuinely useful. Row-level security enables fine-grained access control. ## The Catch The all-in-one convenience creates dependency. Complex queries hit performance limits. Self-hosting requires significant operational knowledge. ## Verdict The Firebase alternative for developers who want real Postgres. --- # Turso > SQLite at the edge with libSQL **Category:** Databases **Pricing:** freemium **URL:** https://turso.tech **GitHub:** https://github.com/tursodatabase/libsql **Added:** 2026-08-19 --- ## The Hook Turso puts SQLite at the edge, globally replicated. ## What's Good SQLite simplicity with global distribution. Embedded replicas mean reads are local. libSQL adds features SQLite lacks. The latency benefits are real for edge-deployed applications. ## The Catch SQLite limitations still apply—no concurrent writes from different regions. The embedded replica model has a learning curve. Less battle-tested than Postgres. ## Verdict SQLite for the edge era, with surprising capabilities. --- # Redis > The in-memory data store everyone uses **Category:** Databases **Pricing:** open-source **URL:** https://redis.io **GitHub:** https://github.com/redis/redis **Added:** 2026-08-19 --- ## The Hook Redis is the cache, queue, and session store you already need. ## What's Good Blazing fast for what it does. Data structures beyond simple key-value unlock powerful patterns. Pub/sub enables real-time features. The ecosystem is massive—every language has a client. ## The Catch Memory-bound storage is expensive at scale. Persistence options have tradeoffs. The licensing changes caused uncertainty. Clustering adds complexity. ## Verdict The default choice for caching and session storage. --- # Upstash > Serverless Redis and Kafka **Category:** Databases **Pricing:** freemium **URL:** https://upstash.com **Added:** 2026-08-19 --- ## The Hook Upstash is Redis that scales to zero and bills per request. ## What's Good HTTP-based access works from serverless functions without connection management. Pay-per-request pricing makes small projects cheap. Global replication reduces latency. QStash adds serverless-friendly job queues. ## The Catch HTTP adds latency compared to native Redis protocol. High-throughput workloads get expensive. Some Redis features are not available. ## Verdict Redis that fits the serverless model perfectly. --- # Pinecone > Vector database for AI applications **Category:** Databases **Pricing:** freemium **URL:** https://pinecone.io **Added:** 2026-08-19 --- ## The Hook Pinecone is where your embeddings live. ## What's Good Purpose-built for vector similarity search. Managed infrastructure handles scaling. Metadata filtering enables hybrid search. The free tier is generous for experimentation. ## The Catch Vendor lock-in for a core AI capability. Costs scale with vector count and queries. Self-hosting is not an option. ## Verdict The managed vector database for production AI. --- # Qdrant > Open-source vector database with filtering **Category:** Databases **Pricing:** open-source **URL:** https://qdrant.tech **GitHub:** https://github.com/qdrant/qdrant **Added:** 2026-08-19 --- ## The Hook Qdrant is the vector database you can actually self-host. ## What's Good Rust-based performance is excellent. Advanced filtering capabilities enable complex queries. The managed cloud option exists when you need it. Self-hosting keeps costs predictable. ## The Catch Operational burden of self-hosting. Smaller ecosystem than Pinecone. Some advanced features require understanding internals. ## Verdict Self-hostable vector search with production-grade features. --- # ClickHouse > Column-oriented database for analytics **Category:** Databases **Pricing:** open-source **URL:** https://clickhouse.com **GitHub:** https://github.com/ClickHouse/ClickHouse **Added:** 2026-08-19 --- ## The Hook ClickHouse makes billion-row queries feel instant. ## What's Good Analytics queries that would take minutes in Postgres complete in seconds. Column-oriented storage is perfect for time-series and event data. Compression is aggressive. The SQL dialect is familiar. ## The Catch Not a general-purpose database—updates and deletes are expensive. Operational complexity for self-hosting. The learning curve for optimal schema design is steep. ## Verdict The analytics database when Postgres is too slow. --- # Timescale > Time-series superpowers for Postgres **Category:** Databases **Pricing:** freemium **URL:** https://timescale.com **GitHub:** https://github.com/timescale/timescaledb **Added:** 2026-08-19 --- ## The Hook Timescale is Postgres that handles time-series at scale. ## What's Good Full Postgres compatibility means your existing tools work. Automatic partitioning handles time-series data at scale. Continuous aggregates precompute expensive queries. The managed platform handles operations. ## The Catch Adds complexity to Postgres operations. Some hypertable features have limitations. The learning curve for optimal performance is real. ## Verdict Time-series data without leaving Postgres. --- # DuckDB > In-process OLAP database **Category:** Databases **Pricing:** open-source **URL:** https://duckdb.org **GitHub:** https://github.com/duckdb/duckdb **Added:** 2026-08-19 --- ## The Hook DuckDB is SQLite for analytics—fast, embedded, and serverless. ## What's Good Analytics queries run locally without a server. Parquet, CSV, and JSON import just works. The SQL dialect is comprehensive. Integration with Python and R is seamless. ## The Catch Not for OLTP workloads—it is analytics-focused. Single-node limits apply. Less ecosystem than traditional databases. ## Verdict Local analytics that just works. --- # SQLite > The most deployed database in the world **Category:** Databases **Pricing:** open-source **URL:** https://sqlite.org **Added:** 2026-08-19 --- ## The Hook SQLite is good enough for more than you think. ## What's Good Zero configuration, zero server, zero administration. Single-file database is infinitely portable. Write-ahead logging handles concurrent reads. Used by every browser and mobile app. ## The Catch Single writer at a time limits write concurrency. Not designed for network access. Feature set is smaller than Postgres. ## Verdict Good enough for more applications than you realize. --- # PocketBase > SQLite backend, auth, and admin UI in one Go binary **Category:** Databases **Pricing:** open-source **URL:** https://pocketbase.io **GitHub:** https://github.com/pocketbase/pocketbase **Added:** 2026-09-03 --- ## The Hook PocketBase is Firebase if Firebase were one file you scp to a $5 VPS. ## What's Good MIT. Realtime, auth, file storage, and an admin UI from a single binary. Embed it as a Go library or run pocketbase serve. JS client talks REST. This is the backend you hand to a friend without a cloud bill. ## The Catch SQLite is the database. Concurrent writes and multi-region are not the product. There is no official hosted PocketBase Cloud. You run the binary, you own backups, you own the disk filling up. ## Verdict The one-binary BaaS. Perfect until you need Postgres. --- # TigerBeetle > Apache-2.0 financial-transactions database, not general OLTP **Category:** Databases **Pricing:** open-source **URL:** https://tigerbeetle.com **GitHub:** https://github.com/tigerbeetle/tigerbeetle **Added:** 2026-09-03 --- ## The Hook TigerBeetle is a database for money, with a safety model Postgres was not designed for. ## What's Good Apache-2.0. Debit/credit double-entry, strict serializability, replica safety. Single static binary. The docs are explicit: ledgers, not general CRUD. ## The Catch Not a Postgres replacement. No ad-hoc SQL, no JSON documents. Clustered replicas. If you needed a general database, this is the wrong gem. ## Verdict The financial transactions database. Apache-2.0, specialized on purpose. --- # ParadeDB > AGPL-3.0 Postgres search + analytics; Community free forever **Category:** Databases **Pricing:** open-source **URL:** https://www.paradedb.com **GitHub:** https://github.com/paradedb/paradedb **Added:** 2026-09-04 --- ## The Hook ParadeDB puts search inside Postgres when a separate search server is the wrong box. ## What's Good AGPL-3.0 Community. Self-managed free forever on a single node. pg_search for full-text and hybrid search plus analytics on Postgres. Verified homepage/pricing 2026-09-04. ## The Catch Search + analytics inside Postgres (pg_search) — not a separate search server. Distinct from Typesense/Meilisearch (cut: no free forever Cloud). Loud: Cloud is waitlist/beta — NOT a free hosted tier. Enterprise HA is sales; do not reprint Enterprise. ## Verdict AGPL Postgres search. Community self-host free forever. Cloud is not the free path. --- # Valkey > BSD-3-Clause forever-OSS Redis-protocol KV after the license fork **Category:** Databases **Pricing:** open-source **URL:** https://valkey.io **GitHub:** https://github.com/valkey-io/valkey **Added:** 2026-09-04 --- ## The Hook Valkey is the Redis-protocol store you pick when forever-OSS is the requirement, not a speed claim. ## What's Good BSD-3-Clause. Linux Foundation project after the Redis license change. Drop-in Redis protocol clients and data structures. Self-host is the free path — no Cloud SKU invented here. ## The Catch Forever-OSS Redis-protocol fork lane, not Redis but faster. Redis gem is already live — this is the license/governance fork, not a performance twin. Do not invent managed free tiers. ## Verdict BSD-3 forever-OSS Redis protocol. Fork lane, not a speed upgrade. --- # Litestream > Apache-2.0 SQLite WAL replication to S3 or local disk **Category:** Databases **Pricing:** open-source **URL:** https://litestream.io **GitHub:** https://github.com/benbjohnson/litestream **Added:** 2026-09-04 --- ## The Hook Litestream is the WAL stream that makes a single SQLite file survivable off-box. ## What's Good Apache-2.0. Continuous SQLite WAL replication to S3-compatible object storage or local paths. Restore by pointing at a replica. Ops binary, not a new query engine. ## The Catch Ops/replication beside SQLite/Turso/PocketBase — not a new database. You still pick the SQLite host. Distinct from Turso (libSQL edge) and PocketBase (BaaS binary). ## Verdict Apache-2.0 SQLite WAL stream. Replication tool, not a DB product. --- # Atlas > Apache-2.0 schema-as-code CLI; Cloud CI/governance is paid **Category:** Databases **Pricing:** freemium **URL:** https://atlasgo.io **GitHub:** https://github.com/ariga/atlas **Added:** 2026-09-04 --- ## The Hook Atlas is Terraform-for-schemas when hand-written migrate files or ORM-only diffs are the pain. ## What's Good Apache-2.0 CLI forever: declarative schema apply and versioned migrate diff/lint/apply. HCL, SQL, or ORM loaders. Local linting with a disposable --dev-url. No per-migration license tax on the CLI. ## The Catch Vs Flyway/Liquibase/Prisma migrate: schema-as-code + auto-planned diffs, not only ordered SQL scripts. LOUD: Atlas Cloud (CI automation, drift, governance) is a separate paid product — Starter free-limited, Pro adds per-dev/project fees. Free path is the OSS CLI. ## Verdict Apache-2.0 schema CLI. Cloud CI/governance is paid; CLI is the free path. --- # Electric > Apache-2.0 Postgres read-path sync; Cloud PAYG under $5 waived **Category:** Databases **Pricing:** freemium **URL:** https://electric.ax **GitHub:** https://github.com/electric-sql/electric **Added:** 2026-09-04 --- ## The Hook Electric is Postgres→clients sync over HTTP shapes when you refuse to invent your own CDC fan-out. ## What's Good Apache-2.0 sync engine. Logical replication in, Shape partial replication out, CDN-friendly HTTP. Writes stay on your existing API. Self-host Docker; Cloud PAYG waives bills under $5/mo (~5M writes). ## The Catch Read-path sync — not a full offline-write CRDT stack or a replacement for your mutation API. Vs PowerSync/Replicache: Postgres-native shapes + free reads on Cloud, not mobile SQLite product packaging. Pro Cloud starts $249/mo. ## Verdict Apache-2.0 Postgres sync. Self-host free; Cloud hobby under $5 waived. --- # Meilisearch > MIT typo-tolerant search engine; Cloud managed, EE sharding paid **Category:** Databases **Pricing:** freemium **URL:** https://www.meilisearch.com **GitHub:** https://github.com/meilisearch/meilisearch **Added:** 2026-09-04 --- ## The Hook Meilisearch is the single-binary search engine when Elasticsearch ops are the complaint and SQL LIKE is a joke. ## What's Good MIT Community Edition. Fast typo-tolerant full-text, filters, faceting, hybrid/AI search path. One binary, great DX, SDKs everywhere. Self-host free for production CE. ## The Catch Vs Typesense/ParadeDB: dedicated search engine UX, not Postgres extension search. LOUD: Enterprise Edition sharding/advanced ops is BUSL — commercial license for prod EE. Cloud is managed paid. CE is the free production path. ## Verdict MIT search CE. Self-host free; EE sharding and Cloud are paid. --- # Bytebase > MIT-core DB change governance; Community free to 20 users/10 instances **Category:** Databases **Pricing:** freemium **URL:** https://www.bytebase.com **GitHub:** https://github.com/bytebase/bytebase **Added:** 2026-09-04 --- ## The Hook Bytebase is the PR-for-schema control plane when Atlas is the schema CLI and Flyway is the changelog runner. ## What's Good MIT-core open-core. Change requests, 200+ SQL review rules, GitOps schema versioning, web SQL editor, multi-engine (Postgres/MySQL/SQL Server/Oracle/Mongo/Redis/Snowflake/…). Community free forever: 20 users, 10 instances — change management included. MCP server ships on Community. ## The Catch Vs Atlas/Liquibase: governance UI + access/review workflow, not only schema-as-code CLI. LOUD: enterprise/ paths are commercial; SSO beyond Google/GitHub, unlimited audit, JIT access, dynamic masking, custom approvals are Pro/Enterprise. Pro $20/user/mo; Community caps users/instances. ## Verdict MIT-core DB DevSecOps. Community free (20/10); Pro/Enterprise for SSO/masking/scale. --- # dbmate > MIT framework-agnostic SQL migration CLI; single Go binary **Category:** Databases **Pricing:** open-source **URL:** https://github.com/amacneil/dbmate **GitHub:** https://github.com/amacneil/dbmate **Added:** 2026-09-04 --- ## The Hook dbmate is plain-SQL migrations in one binary when ORM migrate and Atlas schema-as-code are both heavier than you need. ## What's Good MIT. Language-agnostic up/down/status/dump for Postgres, MySQL, SQLite, ClickHouse. Timestamped SQL files, schema dump, Docker-friendly. No SaaS seat tax. ## The Catch Vs Atlas/Flyway/Bytebase: ordered SQL files only — no declarative planner, no review UI. You own ordering and rollbacks. Distinct from goose-style twins by DX focus plus dump. ## Verdict MIT SQL migration CLI. Tiny free forever; not a schema platform. --- # sqlc > MIT type-safe code gen from SQL **Category:** Databases **Pricing:** open-source **URL:** https://sqlc.dev **GitHub:** https://github.com/sqlc-dev/sqlc **Added:** 2026-09-06 --- ## The Hook sqlc turns annotated SQL into typed query code when you want SQL as source of truth, not an ORM DSL. ## What's Good MIT. Write SQL, get Go/Kotlin/Python/TypeScript stubs with compile-time types. No runtime ORM magic. Great for Postgres/MySQL/SQLite shops that like queries readable. ## The Catch Codegen from SQL - not Prisma/Drizzle ORM, not a database. Schema migrations are still yours (golang-migrate, Atlas, etc.). Language support depth varies. ## Verdict MIT SQL codegen. Free forever; not an ORM. --- # ⚙ DevOps Ship with confidence: CI/CD, containers, infrastructure as code, and deployment automation. *34 gems in this category* --- # GitHub Actions > CI/CD that lives where your code does **Category:** DevOps **Pricing:** freemium **URL:** https://github.com/features/actions **Added:** 2026-08-19 --- ## The Hook GitHub Actions is CI/CD without leaving GitHub. ## What's Good Integration with GitHub is seamless—triggers, permissions, and secrets just work. The marketplace has actions for almost everything. Matrix builds parallelize testing. Self-hosted runners give you control when needed. ## The Catch YAML syntax has a learning curve. Debugging failed workflows is frustrating. Costs scale with build minutes. Complex workflows become hard to maintain. ## Verdict The default CI/CD for GitHub repositories. --- # Docker > Container runtime that changed deployment **Category:** DevOps **Pricing:** freemium **URL:** https://docker.com **GitHub:** https://github.com/docker/docker-ce **Added:** 2026-08-19 --- ## The Hook Docker made "works on my machine" a solvable problem. ## What's Good Containers are the universal deployment unit. Dockerfile syntax is approachable. Docker Compose handles multi-container development. The ecosystem is massive. ## The Catch Resource overhead compared to native. Security requires attention. Desktop licensing changes affected teams. Image size management is ongoing work. ## Verdict Containers are table stakes; Docker is how most start. --- # Kubernetes > Container orchestration at scale **Category:** DevOps **Pricing:** open-source **URL:** https://kubernetes.io **GitHub:** https://github.com/kubernetes/kubernetes **Added:** 2026-08-19 --- ## The Hook Kubernetes is what you need when Docker Compose is not enough. ## What's Good Declarative infrastructure scales to massive deployments. Self-healing restarts failed containers. Rolling updates minimize downtime. The ecosystem of operators extends capabilities. ## The Catch Complexity is legendary. Small teams often do not need it. Operational burden is significant. The YAML can become unmanageable. ## Verdict Powerful but only worth it at scale. --- # Terraform > Infrastructure as code that actually works **Category:** DevOps **Pricing:** open-source **URL:** https://terraform.io **GitHub:** https://github.com/hashicorp/terraform **Added:** 2026-08-19 --- ## The Hook Terraform turns cloud infrastructure into version-controlled code. ## What's Good Multi-cloud support with one tool. State management tracks what exists. Plan before apply prevents surprises. The provider ecosystem covers almost everything. ## The Catch State management requires care. The licensing change to BSL concerned some users. HCL has quirks. Complex modules become hard to debug. ## Verdict The standard for infrastructure as code. --- # Pulumi > Infrastructure as code in real languages **Category:** DevOps **Pricing:** freemium **URL:** https://pulumi.com **GitHub:** https://github.com/pulumi/pulumi **Added:** 2026-08-19 --- ## The Hook Pulumi lets you write infrastructure in TypeScript, Python, or Go. ## What's Good Real programming languages mean real abstractions—loops, conditionals, functions. Testing infrastructure becomes possible. IDE support is excellent. Type safety catches errors early. ## The Catch Smaller ecosystem than Terraform. Learning curve for infrastructure patterns. The state service is a dependency unless self-hosted. ## Verdict IaC for developers who want real programming languages. --- # Fly.io > Deploy apps to VMs close to your users **Category:** DevOps **Pricing:** freemium **URL:** https://fly.io **Added:** 2026-08-19 --- ## The Hook Fly.io gives you VMs at the edge without the edge constraints. ## What's Good Full VMs mean no serverless limitations. Global distribution with anycast routing. Postgres, Redis, and volumes built-in. The CLI is developer-friendly. ## The Catch More operational than PaaS alternatives. Pricing requires understanding. Some features are less polished. The free tier has limitations. ## Verdict Full VMs globally for developers who need more than serverless. --- # Railway > Deploy from GitHub with zero configuration **Category:** DevOps **Pricing:** freemium **URL:** https://railway.app **Added:** 2026-08-19 --- ## The Hook Railway is Heroku for the GitHub generation. ## What's Good Push to deploy works out of the box. Managed Postgres and Redis with one click. The dashboard shows logs and metrics clearly. Preview environments for PRs. ## The Catch Less control than VMs or containers. Pricing scales with usage. Customization options are limited. Some workloads do not fit the model. ## Verdict The simplest path from code to production. --- # Coolify > Self-hosted Heroku alternative **Category:** DevOps **Pricing:** open-source **URL:** https://coolify.io **GitHub:** https://github.com/coollabsio/coolify **Added:** 2026-08-19 --- ## The Hook Coolify is the self-hosted platform you can actually run. ## What's Good Full control on your own infrastructure. Deploy any language or framework. Manages databases, Redis, and S3 storage. The UI handles what Docker Compose cannot. ## The Catch Self-hosting means self-operating. Some features are less polished than commercial alternatives. Single-server focus limits scale. ## Verdict Self-hosted PaaS that is actually usable. --- # CapRover > Free PaaS you can run anywhere **Category:** DevOps **Pricing:** open-source **URL:** https://caprover.com **GitHub:** https://github.com/caprover/caprover **Added:** 2026-08-19 --- ## The Hook CapRover turns any VPS into a Heroku-like platform. ## What's Good One-command install on any Docker host. One-click apps from the marketplace. SSL certificates handled automatically. Multi-server clustering works. ## The Catch Documentation assumes Docker knowledge. Some advanced features require manual configuration. Less active development than alternatives. ## Verdict Simple self-hosted PaaS for small deployments. --- # Biome > Rust formatter and linter that replaces Prettier plus ESLint **Category:** DevOps **Pricing:** open-source **URL:** https://biomejs.dev **GitHub:** https://github.com/biomejs/biome **Added:** 2026-09-03 --- ## The Hook Biome is the JS toolchain that finally made format and lint one binary. ## What's Good MIT OR Apache-2.0. Formatter and linter in Rust, one config, one CLI. Rome's successor with a real release cadence. Drop-in enough that many repos delete prettier, eslint, and a handful of plugins on the same afternoon. ## The Catch Not a 1:1 ESLint plugin ecosystem. Some rules and framework plugins do not exist yet. Dual-license is MIT OR Apache-2.0 at your option — there is no single LICENSE file at the repo root. ## Verdict The default JS format+lint binary in 2026. Fast, dual-licensed, not ESLint. --- # Kamal > Deploy Docker containers to your own VMs with one CLI **Category:** DevOps **Pricing:** open-source **URL:** https://kamal-deploy.org **GitHub:** https://github.com/basecamp/kamal **Added:** 2026-09-03 --- ## The Hook Kamal is Capistrano for Docker, from the people who still run their own hardware. ## What's Good MIT (MIT-LICENSE). kamal setup / kamal deploy from a laptop to a VM. Traefik or Kamal Proxy, accessories, secrets, rolling deploys. You keep the servers; you skip Kubernetes. Built at 37signals for HEY and Basecamp. ## The Catch You still run Docker hosts. This is not a PaaS. Networking, disks, and what-if-the-VM-dies remain your problem. Kamal 2 changed proxy defaults — read the current docs, not a 2024 blog post. ## Verdict The honest alternative to a PaaS when you already have VMs. --- # n8n > Fair-code workflow automation you can self-host **Category:** DevOps **Pricing:** source-available **URL:** https://n8n.io **GitHub:** https://github.com/n8n-io/n8n **Added:** 2026-09-03 --- ## The Hook n8n is Zapier you can run yourself — until you try to sell it as a service. ## What's Good Source is public. Self-host for internal business, non-commercial, or personal use. Visual workflows, hundreds of nodes, and a real queue. Cloud exists if you do not want to run Node. Fair-code is the model they named. ## The Catch Sustainable Use License is not OSI. Use and modification: internal / non-commercial / personal. Distribution only free of charge for non-commercial purposes. Files with .ee. in the name need a separate Enterprise License. Do not call it open source. ## Verdict The workflow engine with a catch in the LICENSE.md. SUL, not OSI. --- # SST > MIT infrastructure-as-code for AWS, with live Lambda dev **Category:** DevOps **Pricing:** open-source **URL:** https://sst.dev **GitHub:** https://github.com/sst/sst **Added:** 2026-09-03 --- ## The Hook SST is the AWS toolkit that makes Lambda feel like a local server. ## What's Good MIT. Components for functions, queues, databases, and frontends on AWS. sst dev tails live cloud resources. Ion rewrite is the current line. You keep the AWS account. ## The Catch AWS-shaped. Other clouds are not the happy path. You still pay AWS. The Ion/v3 cut left a trail of outdated v2 tutorials — read current sst.dev docs. ## Verdict MIT IaC for AWS. Live dev is the feature; the bill is still Amazon. --- # Mise > MIT polyglot runtime manager, the asdf successor that is fast **Category:** DevOps **Pricing:** open-source **URL:** https://mise.jdx.dev **GitHub:** https://github.com/jdx/mise **Added:** 2026-09-03 --- ## The Hook Mise is asdf without the wait, plus tasks and env in one TOML file. ## What's Good MIT. Install Node, Python, Ruby, Go, and the rest from one CLI. mise.toml pins versions, env, and tasks. Drop-in asdf plugin compatibility for most tools. Written in Rust by Jeff Dickey (rtx). ## The Catch Not a container runtime and not direnv-only. Some asdf plugins still assume bash. Team-wide adoption means every laptop runs mise, or CI drifts. ## Verdict The version manager to install in 2026. MIT, fast, one toml. --- # Dagger > Apache-2.0 CI as containers, programmable in your language **Category:** DevOps **Pricing:** open-source **URL:** https://dagger.io **GitHub:** https://github.com/dagger/dagger **Added:** 2026-09-03 --- ## The Hook Dagger is CI you can run on a laptop because the graph is containers. ## What's Good Apache-2.0. Write pipelines in Go, Python, or TypeScript. Same graph locally and in GitHub Actions / any Docker host. Caching is content-addressed. Engine is a container runtime, not a YAML dialect. ## The Catch You need Docker or a compatible engine. Rewriting mature CI YAML is a project, not an afternoon. Dagger Cloud is a separate product — this review does not invent its free tier. ## Verdict Apache-2.0 portable CI. The pipeline is code; Cloud is optional and unverified here. --- # Garage > AGPL-3.0 S3-compatible store for small geo-distributed hosts **Category:** DevOps **Pricing:** open-source **URL:** https://garagehq.deuxfleurs.fr **GitHub:** https://github.com/deuxfleurs-org/garage **Added:** 2026-09-04 --- ## The Hook Garage is the S3-compatible binary you run when Ceph or MinIO-scale is the wrong weight. ## What's Good AGPL-3.0. Entirely free self-host. Single binary. Built for small and geo-distributed deployments. No Cloud free-tier product — self-host is the path. ## The Catch AGPL. S3-compatible for small/geo-distributed self-host — not SeaweedFS/Ceph/MinIO-scale. SeaweedFS omitted once this publishes. Stretch: object store outside core remit, kept for the free forever self-host bar. ## Verdict AGPL S3 for small clusters. Self-host free. Not MinIO-scale. --- # Temporal > MIT self-host workflow engine; Cloud is credits then paid **Category:** DevOps **Pricing:** freemium **URL:** https://temporal.io **GitHub:** https://github.com/temporalio/temporal **Added:** 2026-09-04 --- ## The Hook Temporal is durable workflow execution when the job outlives a request and a retry queue. ## What's Good Service MIT. Self-host the Temporal Service forever at $0. SDKs, workflows, activities, visibility. Skills already on the catalog; gem was missing. ## The Catch Not Inngest/Trigger.dev — different durable-execution shape. Cloud is credits-then-paid ($1k/90d style credits, then paid) — never free forever. Local temporal server start-dev is not production. OSS self-host is the $0 hatch. ## Verdict MIT self-host durable workflows. Cloud is not a free forever hatch. --- # Caddy > Apache-2.0 reverse proxy with automatic HTTPS by default **Category:** DevOps **Pricing:** open-source **URL:** https://caddyserver.com **GitHub:** https://github.com/caddyserver/caddy **Added:** 2026-09-04 --- ## The Hook Caddy is the reverse proxy that gets TLS right before you open a config file. ## What's Good Apache-2.0. Automatic HTTPS (Let us Encrypt / ZeroSSL) by default. Reverse proxy, static file server, extensible with modules. One binary. Coolify/CapRover assume a proxy — this is often that proxy. ## The Catch Encyclopedia-adjacent but was absent while Coolify/CapRover assume a proxy. Caddy is the proxy, not the PaaS. Not nginx with a friendlier name — ACME-first defaults are the product. ## Verdict Apache-2.0 auto-HTTPS proxy. The missing piece under self-host PaaS. --- # OpenTofu > MPL-2.0 Linux Foundation Terraform fork; truly OSS IaC **Category:** DevOps **Pricing:** open-source **URL:** https://opentofu.org **GitHub:** https://github.com/opentofu/opentofu **Added:** 2026-09-04 --- ## The Hook OpenTofu is the MPL drop-in when Terraform BSL license risk is the actual procurement problem. ## What's Good MPL-2.0 under Linux Foundation / CNCF gravity. HCL-compatible with most Terraform providers/modules. Native state encryption shipped; community registry. Swap binary, tofu init — usually. ## The Catch Vs Terraform/Pulumi: license-first fork of Terraform 1.5-era lineage — not Pulumi programming-language IaC. Not HashiCorp Cloud/Enterprise features. Provider edge cases exist; verify before big migrations. ## Verdict MPL-2.0 Terraform-compatible IaC. The OSS license answer to BSL. --- # Headscale > BSD-3 self-hosted Tailscale control server **Category:** DevOps **Pricing:** open-source **URL:** https://headscale.net **GitHub:** https://github.com/juanfont/headscale **Added:** 2026-09-04 --- ## The Hook Headscale is the control plane you run when Tailscale clients are fine but Tailscale Inc coordination is not. ## What's Good BSD-3-Clause. Implements Tailscale control-server API. Official Tailscale clients talk to your coordination server. ACLs, tags, ephemeral nodes for CI. Small Go binary ops footprint. ## The Catch Control server only — not a userspace WireGuard client replacement. Not associated with Tailscale Inc as a product. You operate DERP/coordination yourself; MagicDNS/feature parity lags Tailscale SaaS at the edges. ## Verdict BSD-3 Tailscale-compatible control plane. Self-host; clients stay Tailscale. --- # Woodpecker CI > Apache-2.0 container-native CI; self-host forever, no paid tier **Category:** DevOps **Pricing:** open-source **URL:** https://woodpecker-ci.org **GitHub:** https://github.com/woodpecker-ci/woodpecker **Added:** 2026-09-04 --- ## The Hook Woodpecker is the Drone-shaped CI you keep when Harness licensing is the reason you left. ## What's Good Apache-2.0. Pipeline-as-YAML, each step a container. Agents you scale. Works with GitHub/GitLab/Gitea/Forgejo/Bitbucket. No feature gate, no seat tax — software is free. ## The Catch Self-host only — not GitHub Actions managed UX. Smaller plugin ecosystem than Actions/Drone-era commercial. You own runners, secrets, and forge OAuth. Distinct from GitHub Actions and Buildkite hosted. ## Verdict Apache-2.0 container CI. Self-host free forever; no SaaS tier. --- # Trivy > Apache-2.0 all-in-one vuln/IaC/secret/SBOM scanner; no paid tier **Category:** DevOps **Pricing:** open-source **URL:** https://trivy.dev **GitHub:** https://github.com/aquasecurity/trivy **Added:** 2026-09-04 --- ## The Hook Trivy is the single binary that scans images, repos, IaC, and K8s when stitching four scanners is the tax. ## What's Good Apache-2.0. CVE vuln scan, misconfig, secrets, licenses, SBOM. Targets: containers, filesystems, git, Terraform/CloudFormation/Dockerfiles, Kubernetes. Default scanner in GitLab/Harbor/Artifact Hub. No feature gate on the CLI. ## The Catch CLI/CI scanner — not Aqua Platform management UI/RBAC/policy SaaS. Vs Grype/Snyk: Aqua OSS default with broad target surface; Aqua Inc monetizes a separate commercial platform. You still triage and patch. ## Verdict Apache-2.0 security scanner. Fully free CLI; Aqua Platform is the paid cousin. --- # NetBird > WireGuard mesh ZTNA; Cloud free 5 users/100 machines, self-host AGPL plane **Category:** DevOps **Pricing:** freemium **URL:** https://netbird.io **GitHub:** https://github.com/netbirdio/netbird **Added:** 2026-09-04 --- ## The Hook NetBird is the full WireGuard mesh control plane you can self-host when Headscale is Tailscale-client-only coordination. ## What's Good Clients BSD-3; management/signal/relay AGPLv3. P2P WireGuard mesh, SSO/MFA, access policies, routes, private DNS. Cloud Free: 5 users, 100 machines. Self-host Community unlimited for core stack. ## The Catch Vs Headscale/Tailscale: own clients+plane. LOUD: Cloud Team from about 6 EUR/user; self-host HA/SCIM/device-approval need Commercial Starter or Enterprise. Dual license on control plane. ## Verdict WireGuard ZTNA. Cloud free 5/100; self-host AGPL core; HA extras paid. --- # Cosign > Apache-2.0 Sigstore keyless container/artifact signing CLI **Category:** DevOps **Pricing:** open-source **URL:** https://docs.sigstore.dev/cosign/signing/overview/ **GitHub:** https://github.com/sigstore/cosign **Added:** 2026-09-04 --- ## The Hook Cosign is keyless OCI signing when long-lived Notary keys are the supply-chain liability. ## What's Good Apache-2.0. Sign/verify containers and blobs; store signatures beside digests in the registry. Keyless via Fulcio OIDC + Rekor transparency log. Also key-pair/KMS. CI-native cosign sign/verify. ## The Catch Signing CLI + public-good Sigstore infra — not a full policy admission controller (pair with Kyverno/OPA). Offline/air-gap needs different PKI. Registry must support referrers/OCI artifacts cleanly. ## Verdict Apache-2.0 Sigstore signing. Free CLI; policy engines are separate. --- # Gitleaks > MIT secret scanner for git/dirs/stdin; CLI free, org Action needs free key **Category:** DevOps **Pricing:** open-source **URL:** https://gitleaks.io **GitHub:** https://github.com/gitleaks/gitleaks **Added:** 2026-09-04 --- ## The Hook Gitleaks is the regex-first secrets detector when Trivy secrets are one of many scanners and you want git-history depth. ## What's Good MIT CLI. Scan git history, directories, stdin. Pre-commit hook + GitHub Action. Huge rule set for API keys/tokens. No SaaS seat tax on the binary. ~29k stars, battle-tested in CI. ## The Catch LOUD: maintainer marks feature-complete (security patches; focus shifting to Betterleaks). Official Action for GitHub org repos needs a free license key from gitleaks.io — personal accounts exempt. Not a full ASVS/SAST suite. False positives need allowlists. ## Verdict MIT secrets CLI. Free forever; org Action wants free license key. --- # Grype > Apache-2.0 SBOM-first vulnerability scanner from Anchore; free CLI **Category:** DevOps **Pricing:** open-source **URL:** https://anchore.com/grype/ **GitHub:** https://github.com/anchore/grype **Added:** 2026-09-04 --- ## The Hook Grype is vuln matching against SBOMs when Trivy is the all-in-one and you already emit Syft/SPDX/CycloneDX. ## What's Good Apache-2.0. Scan images, dirs, or SBOMs; pairs with Syft (generate once, rescan often). Offline DB after download. Transparent matching logic. No feature gate on the CLI. ## The Catch Vuln scanner — not Trivy's IaC/secrets swiss-army, not Cosign signing, not Anchore Enterprise governance UI. You still triage CVEs. Anchore monetizes Enterprise separately. ## Verdict Apache-2.0 SBOM vuln scanner. Free CLI; Enterprise is the paid cousin. --- # Traefik > MIT dynamic reverse proxy for Docker/K8s ingress **Category:** DevOps **Pricing:** open-source **URL:** https://traefik.io **GitHub:** https://github.com/traefik/traefik **Added:** 2026-09-05 --- ## The Hook Traefik is the label-driven load balancer when containers appear and disappear faster than you rewrite nginx.conf. ## What's Good MIT Proxy. Native Docker/Kubernetes/Consul providers, automatic service discovery, Let's Encrypt resolvers, middlewares. One binary that reconfigures as your fleet changes. Traefik Labs Hub/Enterprise are optional add-ons — the proxy itself stays free. ## The Catch Vs Caddy: dynamic Docker/K8s LB/Ingress lane, not auto-HTTPS-for-one-binary. Steeper label/static-vs-dynamic model. Not a PaaS. Hub/API-management SKUs are separate paid products you do not need for reverse-proxy duty. ## Verdict MIT dynamic proxy for Docker/K8s. Free core; Hub is optional paid. --- # Lefthook > MIT Go manager for polyglot repo git-hooks **Category:** DevOps **Pricing:** open-source **URL:** https://lefthook.dev **GitHub:** https://github.com/evilmartians/lefthook **Added:** 2026-09-05 --- ## The Hook Lefthook is fast parallel pre-commit automation when your monorepo is not a Node-only Husky shop. ## What's Good MIT Go binary. One YAML config, parallel jobs, works across Node/Ruby/Python. Single dependency-free binary. ## The Catch Vs Husky: polyglot hooks; not a CI product. Local only. Teams still need a shared install path. ## Verdict MIT polyglot hooks. Free CLI; not CI. --- # just > CC0-1.0 command runner — modern Makefile replacement **Category:** DevOps **Pricing:** open-source **URL:** https://just.systems **GitHub:** https://github.com/casey/just **Added:** 2026-09-05 --- ## The Hook just is the project command runner when Make build-system brain tax is the thing you are trying to delete. ## What's Good CC0-1.0 (public domain dedication). justfile recipes with clear errors, args, .env loading, modules. Not a build graph — a command runner. Cross-platform binary, huge community. ## The Catch Modern Makefile replacement; not Taskfile this wave. No implicit build DAG like Make. You still pick how CI invokes recipes. Distinct from package.json scripts if the repo is polyglot. ## Verdict CC0 command runner. Free forever; Taskfile is a different gem later. --- # Flux > Apache-2.0 CNCF GitOps Toolkit for Kubernetes **Category:** DevOps **Pricing:** open-source **URL:** https://fluxcd.io **GitHub:** https://github.com/fluxcd/flux2 **Added:** 2026-09-05 --- ## The Hook Flux is the CNCF GitOps reconciler when your source of truth is a git repo and the cluster must converge itself. ## What's Good Apache-2.0. GitOps Toolkit controllers (source, kustomize, helm, notification, image automation). Continuous reconciliation, multi-tenancy patterns. Skills already on the catalog; gem was missing. Self-host at zero cost on your cluster. ## The Catch K8s-only GitOps Toolkit; not Kamal/Coolify VM PaaS. You need a working cluster plus RBAC plus secrets story. Prefer current Flux Operator docs over legacy bootstrap blogs. ## Verdict Apache-2.0 K8s GitOps. Free on-cluster; not a VM deploy tool. --- # lazygit > MIT Go terminal UI for git **Category:** DevOps **Pricing:** open-source **URL:** https://github.com/jesseduffield/lazygit **GitHub:** https://github.com/jesseduffield/lazygit **Added:** 2026-09-05 --- ## The Hook lazygit is the Go git TUI when staging hunks in a terminal is faster than fighting a GUI. ## What's Good MIT Go binary. Interactive staging, commits, branches, rebases in one TUI. Cross-platform, keyboard-driven, huge stars for a reason. ## The Catch Go git TUI — not a GitHub app, not magit (Emacs), not CI. You still need git installed and a terminal habit. ## Verdict MIT git TUI. Free forever; terminal-native git UX. --- # Argo CD > Apache-2.0 CNCF declarative GitOps CD for Kubernetes **Category:** DevOps **Pricing:** open-source **URL:** https://argo-cd.readthedocs.io **GitHub:** https://github.com/argoproj/argo-cd **Added:** 2026-09-06 --- ## The Hook Argo CD is the app-centric GitOps CD UI when your team wants to sync Kubernetes apps from git with a dashboard. ## What's Good Apache-2.0 CNCF. Declarative sync from git to cluster, multi-cluster, SSO, RBAC, ApplicationSets. Strong UI for app health and diffs. Self-host on your cluster at zero license cost. ## The Catch Vs Flux: UI/app-centric GitOps CD lane, not Flux Toolkit. Still K8s-only. You run the control plane, secrets, and promotion story. Not Kamal/Coolify for VMs. ## Verdict Apache-2.0 GitOps CD. Free on-cluster; Flux is the toolkit cousin. --- # Helm > Apache-2.0 Kubernetes package manager **Category:** DevOps **Pricing:** open-source **URL:** https://helm.sh **GitHub:** https://github.com/helm/helm **Added:** 2026-09-06 --- ## The Hook Helm is the chart package manager when you want to install and upgrade Kubernetes apps as versioned releases, not raw YAML sprawl. ## What's Good Apache-2.0 CNCF. Charts, values overrides, release history, rollback. De-facto packaging lane for most K8s apps. Huge public chart ecosystem and OCI chart support. ## The Catch K8s package manager - not Argo CD/Flux GitOps sync, not a cluster itself. Templating can become spaghetti; you still own values, secrets, and upgrade discipline. Not Kamal/Coolify for VMs. ## Verdict Apache-2.0 K8s packaging. Free forever; GitOps tools are a different layer. --- # k9s > Apache-2.0 terminal UI for Kubernetes **Category:** DevOps **Pricing:** open-source **URL:** https://k9scli.io **GitHub:** https://github.com/derailed/k9s **Added:** 2026-09-06 --- ## The Hook k9s is the Kubernetes TUI when kubectl get/describe loops are slower than navigating pods live. ## What's Good Apache-2.0 Go binary. Live views of pods, logs, resources, skins, plugins. Keyboard-first cluster ops without leaving the terminal. ## The Catch K8s cluster TUI - not Helm packaging, not a dashboard SaaS. Needs a working kubeconfig and cluster access. Not Lens Electron GUI. ## Verdict Apache-2.0 K8s TUI. Free forever; you still bring the cluster. --- # ◐ Design Build beautiful: component libraries, design systems, CSS frameworks, and dev-designer collaboration. *11 gems in this category* --- # Tailwind CSS > Utility-first CSS that developers love **Category:** Design **Pricing:** open-source **URL:** https://tailwindcss.com **GitHub:** https://github.com/tailwindlabs/tailwindcss **Added:** 2026-08-19 --- ## The Hook Tailwind is CSS for developers who hate CSS. ## What's Good Utility classes eliminate the naming problem. The design system is built-in. JIT compilation means only used styles ship. The plugin ecosystem extends capabilities. ## The Catch HTML becomes verbose. The utility-first paradigm requires adjustment. Complex components need extraction. Some designers find it limiting. ## Verdict The CSS framework that made styling productive. --- # shadcn/ui > Copy-paste components built on Radix **Category:** Design **Pricing:** open-source **URL:** https://ui.shadcn.com **GitHub:** https://github.com/shadcn-ui/ui **Added:** 2026-08-19 --- ## The Hook shadcn/ui is the component library you own, not install. ## What's Good Components copy into your codebase—no dependency to update. Built on Radix primitives for accessibility. Tailwind styling is customizable. The CLI adds components on demand. ## The Catch You own the code, including the maintenance. Updates require manual copying. The paradigm differs from traditional component libraries. ## Verdict The component approach that makes customization natural. --- # Radix UI > Unstyled, accessible React components **Category:** Design **Pricing:** open-source **URL:** https://radix-ui.com **GitHub:** https://github.com/radix-ui/primitives **Added:** 2026-08-19 --- ## The Hook Radix handles accessibility so you can focus on styling. ## What's Good Accessibility is baked in—keyboard navigation, ARIA attributes, focus management. Unstyled means complete design freedom. The API is well-thought-out. Primitives compose well. ## The Catch Styling from scratch takes time. Learning the API requires investment. Some components have quirks. ## Verdict The accessibility foundation for custom design systems. --- # Framer Motion > Production-ready React animations **Category:** Design **Pricing:** open-source **URL:** https://motion.dev **GitHub:** https://github.com/framer/motion **Added:** 2026-08-19 --- ## The Hook Framer Motion makes React animations feel natural. ## What's Good The declarative API matches React mental models. Layout animations work automatically. Gestures are built-in. Exit animations actually work. The spring physics feel right. ## The Catch Bundle size is not negligible. Complex animations require understanding the system. Performance needs attention on mobile. ## Verdict The animation library that makes movement feel right. --- # Figma > Collaborative design tool in the browser **Category:** Design **Pricing:** freemium **URL:** https://figma.com **Added:** 2026-08-19 --- ## The Hook Figma is the browser design file the industry standardized on — and a subscription that adds up. ## What's Good Real-time collaboration changed design workflows. Browser-based means instant access. Developer handoff with inspect mode. Components and variants enable design systems. The plugin ecosystem extends capabilities. ## The Catch Subscription costs add up. Adobe acquisition raised concerns. Offline access is limited. Large files can lag. ## Verdict Default for product design. Budget the seats; it is not a free Figma clone. --- # Storybook > UI component workshop **Category:** Design **Pricing:** open-source **URL:** https://storybook.js.org **GitHub:** https://github.com/storybookjs/storybook **Added:** 2026-08-19 --- ## The Hook Storybook is where you develop components in isolation. ## What's Good Component development outside the app context speeds iteration. Documentation generates automatically. Visual testing catches regressions. Addons extend functionality. ## The Catch Configuration complexity grows. Build times can be slow. Keeping stories updated requires discipline. ## Verdict Essential for component-driven development. --- # Lucide > Beautiful open-source icons **Category:** Design **Pricing:** open-source **URL:** https://lucide.dev **GitHub:** https://github.com/lucide-icons/lucide **Added:** 2026-08-19 --- ## The Hook Lucide is the Feather Icons successor you need. ## What's Good Consistent style across 1500+ icons. Tree-shakeable React components. SVG-based means infinite scaling. Active maintenance adds new icons regularly. ## The Catch Style might not fit every project. Some niches have gaps. Customization requires understanding SVG. ## Verdict The default icon set for modern interfaces. --- # Panda CSS > CSS-in-JS with build-time extraction **Category:** Design **Pricing:** open-source **URL:** https://panda-css.com **GitHub:** https://github.com/chakra-ui/panda **Added:** 2026-08-19 --- ## The Hook Panda CSS is CSS-in-JS without the runtime cost. ## What's Good Build-time extraction means zero runtime. Type-safe tokens catch errors. Recipes enable component variants. RSC compatible out of the box. ## The Catch Different mental model from both Tailwind and traditional CSS-in-JS. Ecosystem is younger. Migration from other solutions takes work. ## Verdict CSS-in-JS for the RSC era. --- # Excalidraw > MIT virtual whiteboard with the hand-drawn look **Category:** Design **Pricing:** open-source **URL:** https://excalidraw.com **GitHub:** https://github.com/excalidraw/excalidraw **Added:** 2026-09-03 --- ## The Hook Excalidraw is the whiteboard architecture reviews actually use. ## What's Good MIT whiteboard you can run in the browser. ## The Catch plus.excalidraw.com is a separate commercial product for teams. This review does not treat Plus as OSS. The OSS app is a drawing surface, not a full design system. ## Verdict The MIT whiteboard. Use the OSS app; Plus is a different SKU. --- # Bento > MIT slide decks as a single HTML file **Category:** Design **Pricing:** open-source **URL:** https://bento.page/slides/ **GitHub:** https://github.com/nyblnet/bento **Added:** 2026-09-03 --- ## The Hook Bento is a slide deck you can email as one HTML file and open offline. ## What's Good MIT (Copyright 2026 The Bento authors). Entire deck in one .bento.html file. Offline. Agent-friendly: the artifact is the file. Bundled reveal.js stays MIT; embedded typefaces are OFL. ## The Catch Reveal.js-based. This is slides-for-agents, not an office suite. Not BentoML and not bentonow.com. ## Verdict One-file MIT slides. Useful, small, not a productivity suite. --- # Penpot > MPL-2.0 Figma-class design; self-host forever + Cloud Professional $0 **Category:** Design **Pricing:** freemium **URL:** https://penpot.app **GitHub:** https://github.com/penpot/penpot **Added:** 2026-09-05 --- ## The Hook Penpot is the serious open-source design tool when Figma seats and Excalidraw whiteboards are the wrong shape of problem. ## What's Good MPL-2.0. Self-host forever with full editor features. Cloud Professional is $0 (team/storage caps: ~8 members, ~10GB, short version history). CSS-aware layouts, components, prototyping, multiplayer. Real design systems lane — not a sticky-note board. ## The Catch Only serious OSS Figma-class design; not Excalidraw whiteboards. LOUD: Cloud Unlimited from ~$7/editor (capped ~$175/mo). Self-host means you own Docker/backups/upgrades. Ecosystem/plugins thinner than Figma marketplace. ## Verdict MPL-2.0 design suite. Self-host free; Cloud Pro $0 then Unlimited ~$7/editor. --- # ◉ Observability See everything: logging, tracing, metrics, error tracking, and the dashboards that make sense of it all. *11 gems in this category* --- # Sentry > Error tracking that actually helps **Category:** Observability **Pricing:** freemium **URL:** https://sentry.io **GitHub:** https://github.com/getsentry/sentry **Added:** 2026-08-19 --- ## The Hook Sentry tells you about errors before users do. ## What's Good Stack traces with source maps make debugging possible. Release tracking ties errors to deploys. Performance monitoring covers web vitals. Session replay shows what users experienced. ## The Catch Event volume drives cost. Self-hosting is complex. Some features are enterprise-gated. Alert noise requires tuning. ## Verdict The error tracking service most teams use. --- # Grafana > Dashboards for everything observable **Category:** Observability **Pricing:** open-source **URL:** https://grafana.com **GitHub:** https://github.com/grafana/grafana **Added:** 2026-08-19 --- ## The Hook Grafana visualizes metrics from anywhere. ## What's Good Connects to every data source imaginable. Dashboard flexibility is unmatched. Alerting rules catch issues. The ecosystem of plugins extends capabilities. ## The Catch Dashboard sprawl becomes a maintenance burden. Learning PromQL or LogQL takes time. Self-hosting requires infrastructure. ## Verdict The universal observability dashboard. --- # Prometheus > Time-series metrics for cloud-native **Category:** Observability **Pricing:** open-source **URL:** https://prometheus.io **GitHub:** https://github.com/prometheus/prometheus **Added:** 2026-08-19 --- ## The Hook Prometheus is how you collect metrics at scale. ## What's Good Pull-based collection simplifies service discovery. PromQL enables powerful queries. The exporters ecosystem covers everything. CNCF graduated project means long-term support. ## The Catch Single-node storage limits scale. PromQL has a learning curve. High cardinality labels cause problems. ## Verdict The foundation of cloud-native metrics. --- # OpenTelemetry > Vendor-neutral observability framework **Category:** Observability **Pricing:** open-source **URL:** https://opentelemetry.io **GitHub:** https://github.com/open-telemetry **Added:** 2026-08-19 --- ## The Hook OpenTelemetry is the observability standard that might actually win. ## What's Good Vendor-neutral means no lock-in. Unified traces, metrics, and logs. Automatic instrumentation for popular frameworks. Growing ecosystem of exporters. ## The Catch The specification is complex. Implementation maturity varies by language. Configuration requires understanding the model. ## Verdict The standard for future-proof instrumentation. --- # Datadog > Unified observability platform **Category:** Observability **Pricing:** paid **URL:** https://datadoghq.com **Added:** 2026-08-19 --- ## The Hook Datadog is the observability platform with everything. ## What's Good Logs, metrics, traces, and APM in one platform. The integrations list is massive. ML-powered anomaly detection helps. The UI is polished. ## The Catch Expensive at scale—costs surprise teams. Vendor lock-in is significant. Feature creep makes the platform complex. ## Verdict The everything platform—if you can afford it. --- # Axiom > Log management without limits **Category:** Observability **Pricing:** freemium **URL:** https://axiom.co **Added:** 2026-08-19 --- ## The Hook Axiom ingests logs without worrying about volume. ## What's Good No per-GB pricing means you can log everything. Query speed is excellent. The APL query language is approachable. Vercel integration works seamlessly. ## The Catch Younger platform than alternatives. Some enterprise features are missing. The ecosystem is smaller. ## Verdict Log management that does not punish you for logging. --- # Better Stack > Uptime monitoring and incident management **Category:** Observability **Pricing:** freemium **URL:** https://betterstack.com **Added:** 2026-08-19 --- ## The Hook Better Stack combines uptime, logs, and incidents. ## What's Good Uptime monitoring from global locations. Status pages look professional. Incident management with on-call scheduling. Logtail handles log aggregation. ## The Catch Multiple products mean multiple things to learn. Pricing across products adds up. Some features overlap with other tools. ## Verdict Unified uptime and incident management. --- # Highlight.io > Open-source session replay and monitoring **Category:** Observability **Pricing:** freemium **URL:** https://highlight.io **GitHub:** https://github.com/highlight/highlight **Added:** 2026-08-19 --- ## The Hook Highlight.io is full-stack observability you can self-host. ## What's Good Session replay shows exactly what happened. Error tracking with context. Log aggregation included. Open-source means self-hosting is possible. ## The Catch Newer than established players. Self-hosting requires significant infrastructure. Some features are less mature. ## Verdict Open-source alternative covering replay, errors, and logs. --- # PostHog > Product analytics, flags, replay, and errors in one MIT-core stack **Category:** Observability **Pricing:** freemium **URL:** https://posthog.com **GitHub:** https://github.com/PostHog/posthog **Added:** 2026-09-03 --- ## The Hook PostHog is the product OS we actually run. Open-core; Cloud is the free path that matches the bar. ## What's Good MIT for everything outside ee/. Cloud README states the first 1 million events, 5k recordings, 1M flag requests, 100k exceptions, and 1500 survey responses are free every month, then usage-based. Same product self-hosted if you want the bill to be your cluster. EU Cloud exists. This is analytics plus flags, replay, surveys, and error tracking — not a page-view pixel. ## The Catch ee/ is a separate proprietary license. posthog-foss strips it if you need a clean MIT tree. Self-hosting is a Postgres + ClickHouse ops job. Cloud is usage-metered after the free allowances; replay is usually the line that bites first. ## Verdict The honest Mixpanel/Amplitude alternative. MIT core, real free Cloud, ee/ is not MIT. --- # SigNoz > MIT-core OTel APM (logs/metrics/traces); Cloud from $49/mo **Category:** Observability **Pricing:** freemium **URL:** https://signoz.io **GitHub:** https://github.com/SigNoz/signoz **Added:** 2026-09-04 --- ## The Hook SigNoz is the OpenTelemetry-native single pane when stitching Grafana+Prometheus+Jaeger is the tax. ## What's Good MIT core Community Edition — self-host APM, logs, traces, metrics, alerts. OTel-native ingestion. Cloud Teams from $49/mo with included usage credit. No per-host gimmick on Cloud usage pricing. ## The Catch Open-core: ee/ enterprise bits (advanced authz/audit-class) are not MIT. Self-host means you operate ClickHouse. Vs Grafana stack/Datadog: all-in-one OTel product, not a la carte LGTM or Datadog SaaS. Cloud is paid after trial. ## Verdict MIT-core OTel observability. Self-host free; Cloud from $49/mo. --- # Uptime Kuma > MIT self-host uptime monitoring and status pages **Category:** Observability **Pricing:** open-source **URL:** https://uptime.kuma.pet **GitHub:** https://github.com/louislam/uptime-kuma **Added:** 2026-09-05 --- ## The Hook Uptime Kuma is the self-host uptime + status page stack when Datadog-class APM is the wrong bill for HTTP checks. ## What's Good MIT. Docker-friendly (commonly on port 3001), monitors, notifications, public status pages. Zero SaaS lock-in for basic uptime. ## The Catch Self-host uptime + status pages — not Datadog/APM. You run the container, backups, and alerting path. HTTP/TCP checks are the lane, not full distributed tracing. ## Verdict MIT self-host uptime. Free forever; you own the box on 3001. --- # ⚡ Rust Systems programming done right: memory safety, performance, and an ecosystem that keeps growing. *14 gems in this category* --- # Rust > Systems programming without the footguns **Category:** Rust **Pricing:** open-source **URL:** https://rust-lang.org **GitHub:** https://github.com/rust-lang/rust **Added:** 2026-08-19 --- ## The Hook Rust is C++ that does not let you shoot yourself. ## What's Good Memory safety without garbage collection. The borrow checker catches bugs at compile time. Performance matches C/C++. The tooling (cargo, rustfmt, clippy) is excellent. ## The Catch The learning curve is steep. Compile times are slow. Fighting the borrow checker is frustrating at first. ## Verdict The systems language for the next decade. --- # Tokio > Async runtime for Rust **Category:** Rust **Pricing:** open-source **URL:** https://tokio.rs **GitHub:** https://github.com/tokio-rs/tokio **Added:** 2026-08-19 --- ## The Hook Tokio is the async foundation everything Rust builds on. ## What's Good Battle-tested at scale by Discord, Cloudflare, and more. The ecosystem (Axum, Tonic, Tower) is comprehensive. Work-stealing scheduler is efficient. The documentation is excellent. ## The Catch Async Rust has a learning curve. Stack traces are confusing. Some patterns from other languages do not translate. ## Verdict The async runtime that powers production Rust. --- # Axum > Web framework from the Tokio team **Category:** Rust **Pricing:** open-source **URL:** https://github.com/tokio-rs/axum **GitHub:** https://github.com/tokio-rs/axum **Added:** 2026-08-19 --- ## The Hook Axum is web development that feels like Rust. ## What's Good Type-safe extractors catch errors at compile time. Tower middleware ecosystem is reusable. No macros means better IDE support. The Tokio team maintains it. ## The Catch Less batteries-included than frameworks like Actix. Error messages can be cryptic. Some patterns require boilerplate. ## Verdict The web framework that embraces Rust idioms. --- # Actix Web > Fast Rust web framework **Category:** Rust **Pricing:** open-source **URL:** https://actix.rs **GitHub:** https://github.com/actix/actix-web **Added:** 2026-08-19 --- ## The Hook Actix Web is raw performance in a web framework. ## What's Good Consistently tops TechEmpower benchmarks. Mature ecosystem with middleware and extensions. The actor model enables certain patterns. Documentation is comprehensive. ## The Catch Macro-heavy API differs from idiomatic Rust. The actor foundation is not always needed. Some community history caused concern. ## Verdict Maximum performance when benchmarks matter. --- # Serde > Serialization framework for Rust **Category:** Rust **Pricing:** open-source **URL:** https://serde.rs **GitHub:** https://github.com/serde-rs/serde **Added:** 2026-08-19 --- ## The Hook Serde makes serialization feel magical. ## What's Good Zero-overhead abstraction over formats. Derive macros minimize boilerplate. Format ecosystem covers JSON, YAML, TOML, and more. Customization is possible when needed. ## The Catch Compile times increase with heavy use. Error messages can be confusing. Some edge cases require manual implementation. ## Verdict The serialization library every Rust project uses. --- # SQLx > Compile-time checked SQL for Rust **Category:** Rust **Pricing:** open-source **URL:** https://github.com/launchbadge/sqlx **GitHub:** https://github.com/launchbadge/sqlx **Added:** 2026-08-19 --- ## The Hook SQLx verifies your queries at compile time. ## What's Good Queries checked against the actual database schema. Async from the ground up. No ORM overhead—just SQL. Supports Postgres, MySQL, and SQLite. ## The Catch Compile-time checking requires database connection. Migration story is simpler than some alternatives. Complex queries need understanding. ## Verdict SQL that the compiler actually checks. --- # Tauri > Build desktop apps with web tech and Rust **Category:** Rust **Pricing:** open-source **URL:** https://tauri.app **GitHub:** https://github.com/tauri-apps/tauri **Added:** 2026-08-19 --- ## The Hook Tauri is Electron without the memory footprint. ## What's Good Bundles are tiny compared to Electron. System webview means no bundled Chromium. Rust backend enables native performance. Security model is well-designed. ## The Catch Webview inconsistencies across platforms. Rust learning curve for backend code. Younger ecosystem than Electron. ## Verdict Desktop apps that do not need 200MB of Chromium. --- # Leptos > Full-stack Rust web framework **Category:** Rust **Pricing:** open-source **URL:** https://leptos.dev **GitHub:** https://github.com/leptos-rs/leptos **Added:** 2026-08-19 --- ## The Hook Leptos is React written in Rust. ## What's Good Fine-grained reactivity without virtual DOM. Server functions blur client/server boundary. WASM output is small. The component model feels familiar to React developers. ## The Catch WASM ecosystem is still maturing. Debugging can be challenging. Smaller community than JavaScript frameworks. ## Verdict Full-stack Rust for the web with a familiar model. --- # clap > Command line argument parser for Rust **Category:** Rust **Pricing:** open-source **URL:** https://docs.rs/clap **GitHub:** https://github.com/clap-rs/clap **Added:** 2026-08-19 --- ## The Hook clap turns structs into command-line interfaces. ## What's Good Derive macros generate parsers from struct definitions. Help text generates automatically. Shell completions work. Subcommands are straightforward. ## The Catch Compile times increase with complex CLIs. Some advanced patterns require builder API. Error messages could be clearer. ## Verdict The CLI argument parser most Rust projects use. --- # Kanidm > MPL-2.0 Rust IdP aiming to replace needing Keycloak **Category:** Rust **Pricing:** open-source **URL:** https://kanidm.com **GitHub:** https://github.com/kanidm/kanidm **Added:** 2026-09-04 --- ## The Hook Kanidm is the Rust identity provider that wants you to stop needing Keycloak. ## What's Good MPL-2.0. Self-host IdP at $0. WebAuthn, OAuth2/OIDC, LDAP, RADIUS, SSH. Written in Rust. Docs and first-party tooling at kanidm.com. ## The Catch Rust IdP aiming to replace needing Keycloak — not the Java CNCF suite. Distinct from Keycloak (Apache-2.0 JVM IAM). You still operate the box. ## Verdict MPL-2.0 Rust IdP. Self-host free. Not Keycloak. --- # Bevy > MIT/Apache Rust ECS engine with first-party Web **Category:** Rust **Pricing:** open-source **URL:** https://bevyengine.org **GitHub:** https://github.com/bevyengine/bevy **Added:** 2026-09-04 --- ## The Hook Bevy is Rust ECS games on the web when Phaser and Pixi stay in JavaScript. ## What's Good MIT OR Apache-2.0. Data-driven ECS, Rust performance, first-party Web (Wasm) path. No royalties. Active engine and ecosystem. ## The Catch Rust ECS + first-party Web — not Phaser/Pixi/Babylon/PlayCanvas JS engines. Steeper Rust learning curve. Distinct from Phaser, Pixi, Babylon, and PlayCanvas. ## Verdict MIT/Apache Rust game engine. Web via Wasm; not a JS canvas framework. --- # Zellij > MIT Rust terminal workspace with layouts and plugins **Category:** Rust **Pricing:** open-source **URL:** https://zellij.dev **GitHub:** https://github.com/zellij-org/zellij **Added:** 2026-09-05 --- ## The Hook Zellij is the Rust terminal workspace agents actually layout when tmux.conf archaeology is the blocker. ## What's Good MIT. Layouts, panes, tabs, WASM plugins, discoverable UI. First-class floating panes and session persistence. Fits the agent-terminal lane: predictable workspaces for long-running coding agents. ## The Catch Rust terminal workspace/layouts/plugins — agent terminal lane; not tmux-clone marketing. Plugin ecosystem younger than tmux. Some SSH/remote habits still assume tmux/screen. ## Verdict MIT Rust terminal workspace. Free forever; layouts over tmux nostalgia. --- # Starship > ISC cross-shell prompt — needs a Nerd Font **Category:** Rust **Pricing:** open-source **URL:** https://starship.rs **GitHub:** https://github.com/starship/starship **Added:** 2026-09-05 --- ## The Hook Starship is the minimal cross-shell prompt when you want git/lang context without writing a theme engine. ## What's Good ISC. Works across shells, fast Rust binary, module config in TOML. Shows git status, language versions, and more with almost no shell glue. ## The Catch Prompt only — not a shell, not Zellij. Needs a Nerd Font for the glyph set. Config is yours to keep sane. ## Verdict ISC prompt. Free forever; font + shell still yours. --- # Helix > MPL-2.0 modal terminal editor — tree-sitter + LSP **Category:** Rust **Pricing:** open-source **URL:** https://helix-editor.com **GitHub:** https://github.com/helix-editor/helix **Added:** 2026-09-06 --- ## The Hook Helix is the modal terminal editor when you want tree-sitter and LSP without a GPU IDE. ## What's Good MPL-2.0 Rust binary. Kakoune/Vim-inspired multiple selections, built-in tree-sitter highlighting, LSP out of the box. No plugin tax to get a modern editing core. Keyboard-first, fast on SSH and local. ## The Catch Vs Zed: terminal modal + tree-sitter/LSP; not a GPU IDE. No Electron GUI, weaker AI-IDE surface than Cursor/Zed. Plugin story is thinner than Neovim. ## Verdict MPL-2.0 terminal editor. Free forever; not Zed/Cursor. --- # ¤ Payments Money that actually moves: CLIs, merchant-of-record, and payment orchestration you can read. *20 gems in this category* --- # Stripe CLI > Official Stripe terminal for webhooks, events, and API objects **Category:** Payments **Pricing:** open-source **URL:** https://docs.stripe.com/stripe-cli **GitHub:** https://github.com/stripe/stripe-cli **Added:** 2026-09-03 --- ## The Hook The Stripe CLI is how webhook development stops being a ngrok folklore. ## What's Good Apache-2.0. stripe listen forwards webhooks to localhost. Event trigger and resend, live request logs, and API object CRUD from the terminal. The binary is free. This is the official tool, not a third-party wrapper. ## The Catch The CLI is free; card charges are not. See stripe.com/pricing for processing rates. Live mode needs a real secret. Dispute fees are on Stripe, not on this binary. ## Verdict Install it. Then still pay Stripe when money moves. --- # Polar > Apache-2.0 merchant-of-record billing for indie and AI products **Category:** Payments **Pricing:** freemium **URL:** https://polar.sh **GitHub:** https://github.com/polarsource/polar **Added:** 2026-09-03 --- ## The Hook Polar is payments for people who do not want to become a tax department. ## What's Good Apache-2.0 (JS SDK is separately MIT). Merchant-of-record for usage, seats, subscriptions, and AI meters. Typed TS/Python SDKs, Next.js and Better Auth adapters, MCP server. Starter is free with no monthly fee: 5 percent plus 50 cents per transaction. polar.sh/pricing 404s; the fees page is the public price list. ## The Catch No first-party CLI in the docs index — SDKs, MCP, and API only. Starter take-rate sits on top of card-brand costs. Polar is MoR, so you do not keep the Stripe account relationship. International cards add extra. Payouts have their own cut. ## Verdict Apache-2.0 MoR for indie and AI products. Free Starter, honest take-rate, real SDKs. --- # Hyperswitch > Apache-2.0 payment orchestrator you can self-host **Category:** Payments **Pricing:** open-source **URL:** https://hyperswitch.io **GitHub:** https://github.com/juspay/hyperswitch **Added:** 2026-09-03 --- ## The Hook Hyperswitch is the payments switch you run when Stripe-as-the-whole-stack is the lock-in. ## What's Good Apache-2.0. Rust orchestrator: routing, vault, retries, 3DS, 90+ connectors. Community Edition is start-building-for-free on the pricing page. About page: free, open-source under Apache 2.0. This is Juspay's control center, open-sourced. ## The Catch Community Edition means you run PCI scope. Cloud and Self-Host Enterprise are Contact Us. The 500+ global enterprises line is Juspay production experience, not a Hyperswitch-customer logo wall. ## Verdict Apache-2.0 payments orchestration. Self-host is free; PCI is not optional. --- # Kill Bill > Apache-2.0 in-house subscriptions and payments, no % take-rate **Category:** Payments **Pricing:** open-source **URL:** https://killbill.io **GitHub:** https://github.com/killbill/killbill **Added:** 2026-09-03 --- ## The Hook Kill Bill is the billing engine you run so nobody takes a cut of your revenue. ## What's Good Apache-2.0. Subscriptions, invoicing, payments, and plugins you host. Self-Hosted OSS is Free — no percentage take-rate. JVM stack, Kaui admin UI, plugin model. You keep the money-movement relationship. ## The Catch JVM and ops-heavy. Aviate is the proprietary accelerator — not the Apache tree. Self-Hosted OSS Free; AWS Marketplace is ~$40/mo plus the AWS bill. No revenue tiles here because there is no % of revenue to quote. ## Verdict Apache-2.0 billing you run. Free until Aviate or the AMI; never a take-rate. --- # Lago > AGPLv3 usage and entitlements metering you self-host **Category:** Payments **Pricing:** open-source **URL:** https://getlago.com **GitHub:** https://github.com/getlago/lago **Added:** 2026-09-03 --- ## The Hook Lago is usage billing that meters events, not a Stripe dashboard with extra fields. ## What's Good AGPLv3. Usage, subscriptions, prepaid, entitlements, invoicing. Self-host the OSS engine. REST API, SDKs, webhooks. Not a payment processor — it talks to Stripe, Adyen, and the rest. ## The Catch Loud: Premium is Contact us. There is no $0 Lago Cloud. Self-host is the free path; Cloud and Premium are quotes. AGPLv3 is network copyleft. ## Verdict AGPLv3 metering. Self-host or talk to sales — there is no hobby cloud. --- # BTCPay Server > MIT self-hosted Bitcoin/LN, 0% fees to your wallet **Category:** Payments **Pricing:** open-source **URL:** https://btcpayserver.org **GitHub:** https://github.com/btcpayserver/btcpayserver **Added:** 2026-09-03 --- ## The Hook BTCPay Server is Bitcoin checkout without a processor taking a cut. ## What's Good MIT. Self-hosted crypto payments to your wallet. On-chain and Lightning (LND, Core Lightning, Eclair). 0% software fees. No KYC, non-custodial. Invoices, POS, donation buttons. You run it or use a shared host. ## The Catch Bitcoin and Lightning to your wallet — not card rails, not a PSP. Distinct from Polar (MoR cards) and Hyperswitch (card orchestration). You operate a node. Network miner fees still exist. ## Verdict MIT Bitcoin/LN checkout. 0% to the software; the node is yours. --- # Medusa > MIT commerce OS you self-host; Cloud from $29/mo **Category:** Payments **Pricing:** open-source **URL:** https://medusajs.com **GitHub:** https://github.com/medusajs/medusa **Added:** 2026-09-03 --- ## The Hook Medusa is the MIT commerce engine you run when Stripe Billing is the wrong shape. ## What's Good MIT. Self-host the commerce OS: products, carts, orders, regions, modules. Pitch is OSS on your infra. Cloud exists and starts From $29/mo (Develop) — verified on medusajs.com/pricing. No GMV tax on the pricing page. ## The Catch Commerce OS, not invoices and not card rails. Cloud From $29/mo is not a $0 hobby hatch — self-host is the free path. Distinct from Polar/Hyperswitch (payments) and from Lago (metering). ## Verdict MIT commerce OS. Self-host is free; Cloud starts at $29/mo. --- # pretix > AGPL-3.0-with-additional-terms event tickets and POS **Category:** Payments **Pricing:** freemium **URL:** https://pretix.eu **GitHub:** https://github.com/pretix/pretix **Added:** 2026-09-03 --- ## The Hook pretix sells conference tickets. It does not sell SaaS subscriptions. ## What's Good AGPL-3.0 with additional terms. Self-host or hosted. Hosted is free for up to 500 free tickets per year. Paid tickets: 2.5% of price excl. tax, never more than EUR 15 per ticket. Shop, access control, and POS are in the product. ## The Catch Loud additional terms: SaaS restrict (do not offer pretix as a service to third parties), footer attribution must stay, modified versions must not look official. Event tickets+POS, not subscriptions. Distinct from Polar (SaaS billing) and Lago (usage metering). ## Verdict Event tickets with a loud AGPL addendum. 500 free tickets/year hosted; 2.5% after that. --- # OpenMeter > Apache-2.0 usage-event metering; Cloud is Kong Konnect now **Category:** Payments **Pricing:** open-source **URL:** https://openmeter.io **GitHub:** https://github.com/openmeterio/openmeter **Added:** 2026-09-03 --- ## The Hook OpenMeter meters usage events first. The hosted twin is no longer a standalone Cloud page. ## What's Good Apache-2.0. Pitch is OSS docker compose: clone, cd quickstart, docker compose up. CloudEvents ingest, meters, entitlements, billing blocks. README: releases are beta and can include breaking changes. ## The Catch Cloud is now Kong Konnect Metering — do not reprint Kong tiles. Distinct from Lago: usage-event metering first, not AGPL usage-billing with no hobby cloud. Compose is a local evaluation stack (Kafka/ClickHouse), not production topology. ## Verdict Apache-2.0 metering you compose locally. Hosted path is Kong; README is beta. --- # Spree Commerce > BSD-3-Clause Ruby commerce OS; Free Sandbox, no platform fees **Category:** Payments **Pricing:** freemium **URL:** https://spreecommerce.org **GitHub:** https://github.com/spree/spree **Added:** 2026-09-04 --- ## The Hook Spree is the Ruby commerce OS you run when Medusa Node stack is the wrong shape. ## What's Good BSD-3-Clause. Self-host OSS with zero platform fees. Free Sandbox (no credit card) verified on the homepage 2026-09-04. REST/OpenAPI, Next.js storefront path, native B2B and multi-vendor. ## The Catch Ruby commerce + REST/OpenAPI + Next.js storefront — not Medusa MIT Node OS. Solidus (BSD sibling) omitted: same Spree-fork slot. Distinct from Medusa. ## Verdict BSD-3 Ruby commerce. Self-host free; Sandbox has no CC. Not Medusa. --- # Sylius > MIT Symfony commerce; Plus is the commercial add-on **Category:** Payments **Pricing:** freemium **URL:** https://sylius.com **GitHub:** https://github.com/Sylius/Sylius **Added:** 2026-09-04 --- ## The Hook Sylius is headless PHP commerce when the stack is Symfony, not Node. ## What's Good MIT. Self-host at $0. Built on Symfony and API Platform. Headless commerce core; Plus is the commercial add-on when you need the paid extras. ## The Catch Symfony + API Platform PHP headless — not Medusa MIT Node OS and not JS. Distinct from Medusa and Spree. Plus is commercial; do not invent rates. ## Verdict MIT PHP commerce on Symfony. Self-host is free; Plus is paid. --- # InvoiceShelf > AGPL-3.0 self-host invoices, estimates, expenses, client portal **Category:** Payments **Pricing:** open-source **URL:** https://invoiceshelf.com **GitHub:** https://github.com/InvoiceShelf/InvoiceShelf **Added:** 2026-09-04 --- ## The Hook InvoiceShelf is Crater continued — SMB invoices on your server, not a billing OS. ## What's Good AGPL-3.0. Self-host everything: invoices, estimates, expenses, client portal, Stripe on your box. No seats, no subscription, no upsell. Stable 2.4.2 (Jul 29 2026). Pitch is OSS on your infra. ## The Catch SMB invoices/estimates/expenses/portal + Stripe on your server — not Kill Bill/Lago subscription orchestration or usage metering. Crater cut re-opened as InvoiceShelf. Distinct from Kill Bill and Lago. ## Verdict AGPL SMB invoicing you host. Free forever; not a metering engine. --- # Bagisto > MIT Laravel/Vue commerce with marketplace and B2B **Category:** Payments **Pricing:** open-source **URL:** https://bagisto.com/en/ **GitHub:** https://github.com/bagisto/bagisto **Added:** 2026-09-04 --- ## The Hook Bagisto is the Laravel/Vue commerce twin when Medusa Node is the wrong stack. ## What's Good MIT. Self-host Laravel + Vue storefront. Marketplace and B2B paths in the OSS tree. No platform fee on the core. ## The Catch Laravel/Vue commerce + marketplace/B2B — PHP twin to Medusa, not Medusa MIT Node OS. Sylius and Spree already cover Symfony/Ruby in Wave 5. Distinct from Medusa. ## Verdict MIT Laravel commerce. Self-host free. Not Medusa Node. --- # PrestaShop > OSL-3.0 classic PHP cart; pitch the core, not the module store **Category:** Payments **Pricing:** open-source **URL:** https://www.prestashop-project.org/ **GitHub:** https://github.com/PrestaShop/PrestaShop **Added:** 2026-09-04 --- ## The Hook PrestaShop is the classic PHP cart when headless Node commerce is overkill. ## What's Good OSL-3.0 core. Self-host the classic catalog/cart/checkout stack. Huge module ecosystem and agency muscle memory. Project site is the OSS pitch. ## The Catch Loud: installer may bundle optional SA commercial modules — pitch core only. Classic PHP cart, not headless Node (Medusa) and not Magento (HOLD). Distinct from Medusa, Bagisto, Sylius, Spree. ## Verdict OSL-3.0 PHP cart. Core is free; ignore the module upsell in the installer. --- # Inkvoice > MIT freelancer invoicing; self-host primary, tiny Cloud Free **Category:** Payments **Pricing:** freemium **URL:** https://inkvoice.app **GitHub:** https://github.com/pigontech/inkvoice **Added:** 2026-09-04 --- ## The Hook Inkvoice is freelancer invoicing you self-host when InvoiceShelf is SMB and Kill Bill is a billing OS. ## What's Good MIT self-host is the primary free path. Cloud Free: 1 user / 5 customers — pitch self-host for real use. Simple invoices for freelancers without Lago metering or Kill Bill orchestration. ## The Catch Freelancer invoicing — not InvoiceShelf SMB suite, not Kill Bill subscription OS, not Lago usage metering. Distinct from InvoiceShelf, Kill Bill, and Lago. ## Verdict MIT freelancer invoices. Self-host free; Cloud Free is a teaser cap. --- # Paymenter > MIT hosting billing; self-host free, cloud PaaS $5/mo paid **Category:** Payments **Pricing:** freemium **URL:** https://paymenter.org **GitHub:** https://github.com/Paymenter/Paymenter **Added:** 2026-09-04 --- ## The Hook Paymenter is the hosting webshop + provisioning panel when Kill Bill or Lago is the wrong shape. ## What's Good MIT. Self-host free forever. Automates subscriptions, invoices, and service deployment for hosting businesses. WHMCS importer. Cloud PaaS starts at $5/mo — paid, not free. ## The Catch Vs Kill Bill/Lago/Polar: hosting webshop + provisioning (Pterodactyl/cPanel, WHMCS importer) — not generic subscription metering or Polar MoR. Cloud is $5/mo paid. Self-host is the $0 path. ## Verdict MIT hosting billing. Self-host free; cloud PaaS is $5/mo paid. --- # SolidInvoice > MIT Symfony invoicing; self-host free, Cloud trial paid **Category:** Payments **Pricing:** freemium **URL:** https://solidinvoice.co/open-source **GitHub:** https://github.com/SolidInvoice/SolidInvoice **Added:** 2026-09-04 --- ## The Hook SolidInvoice is Symfony/PHP freelancer+SMB invoicing when InvoiceShelf is the Laravel cousin. ## What's Good MIT (3.1.x LICENSE). Self-Hosted free forever — solidinvoice.co/open-source confirms $0 software self-host. Quotes→invoices, payments, API/MCP. ## The Catch Vs InvoiceShelf/Inkvoice: Symfony/PHP freelancer+SMB invoicing (quotes→invoices, payments, API/MCP). LOUD: solidinvoice.co leads with 14-day Cloud trial / from $9/mo — free path is self-host ONLY. ## Verdict MIT Symfony invoicing. Self-host $0; Cloud trial/from $9 is paid. --- # Autumn > Apache-2.0 Stripe billing layer; Free to 8K billing volume **Category:** Payments **Pricing:** freemium **URL:** https://useautumn.com **GitHub:** https://github.com/useautumn/autumn **Added:** 2026-09-04 --- ## The Hook Autumn is attach/check/track billing infra on top of Stripe when you are tired of webhook state machines. ## What's Good Apache-2.0. Self-host or Cloud. Model subscriptions, credits, usage limits, entitlements. Three calls: attach (checkout/upgrades), check (access), track (usage). Free Cloud tier: 8K monthly billing volume, 10k customers, 10M API requests. ## The Catch Vs Polar/Lago/OpenMeter: Stripe abstraction + entitlements — not Polar MoR, not Lago/OpenMeter metering-first. Stripe fees still apply. Pro Cloud from $375/mo. Free path is Free tier or self-host. ## Verdict Apache-2.0 Stripe billing layer. Free tier real; still bring Stripe. --- # Paddle CLI > MIT community CLI for full Paddle Billing API; agent-native **Category:** Payments **Pricing:** open-source **URL:** https://omshejul.github.io/paddle-cli/ **GitHub:** https://github.com/omshejul/paddle-cli **Added:** 2026-09-04 --- ## The Hook Paddle CLI is the missing Stripe-CLI-shaped tool for Paddle Billing when the dashboard click-ops is the bottleneck. ## What's Good MIT (omshejul/paddle-cli). Spec-driven against Paddle OpenAPI — products, prices, subs, transactions, adjustments. Keychain/credential-store login, write gating, raw paths for new endpoints. uv/Homebrew install. Agent-friendly inspectable commands. ## The Catch Community CLI — not an official Paddle Inc product; API drift risk. Still need a Paddle merchant account (MoR fees apply). Vs Stripe CLI: Paddle-only. Thin MCP forks exist; this is the general CLI. ## Verdict MIT Paddle Billing CLI. Free tool; Paddle merchant + MoR fees separate. --- # Invoice Ninja > ELv2 source-available invoicing; self-host free, hosted Free 5 clients **Category:** Payments **Pricing:** source-available **URL:** https://www.invoiceninja.com **GitHub:** https://github.com/invoiceninja/invoiceninja **Added:** 2026-09-04 --- ## The Hook Invoice Ninja is freelancer/SMB invoicing you can self-host when SolidInvoice is the PHP twin and Stripe Billing is the subscription engine. ## What's Good ELv2 source-available. Invoices, quotes, payments, expenses, timetracking, client portal. Self-host gets Pro/Enterprise feature set in the code. Hosted Free: 5 clients, unlimited invoices. White-label self-host ~$40/yr. ## The Catch LOUD: NOT OSI open source — ELv2 forbids offering it as competing hosted SaaS. Hosted Pro from ~$14/mo (2026 pricing). Vs Lago/Autumn: client invoicing, not usage metering/MoR. Attribution stays unless white-label. ## Verdict ELv2 invoicing. Self-host free for your biz; hosted Free 5 clients / Pro paid. --- # ◎ Comms Realtime rooms, inboxes, and the sockets that keep people talking. *21 gems in this category* --- # LiveKit > Apache-2.0 realtime rooms, SFU, and agents **Category:** Comms **Pricing:** open-source **URL:** https://livekit.com **GitHub:** https://github.com/livekit/livekit **Added:** 2026-09-03 --- ## The Hook LiveKit is the Apache-2.0 realtime stack OpenAI used for ChatGPT Advanced Voice. ## What's Good Apache-2.0 server and CLI. Rooms, SFU, agents, egress. Self-host the media server — that is the free path. SDKs for web, iOS, Android, and agents. ## The Catch Self-hosting an SFU is ops-heavy. Cloud pricing was unverified, so no Cloud Build $0 claim and no monthly allowance numbers. Do not assume Cloud is free. ## Verdict Apache-2.0 voice/video/agent runtime. Self-host is the free path; Cloud is unverified. --- # Chatwoot > MIT-core support inbox you can self-host for $0/agent **Category:** Comms **Pricing:** freemium **URL:** https://www.chatwoot.com **GitHub:** https://github.com/chatwoot/chatwoot **Added:** 2026-09-03 --- ## The Hook Chatwoot is Intercom-shaped support you can run on your own metal. ## What's Good MIT outside enterprise/. Live chat, email, WhatsApp, social. Self-hosted Community Edition is 0 USD/agent/month, free forever — that is the free path. Cloud Hacker is 0 USD/agent, up to 2 agents, 500 conversations/month, live chat only, 30-day retention. ## The Catch SSO, roles, Captain AI, and voice live in paid/enterprise directories — MIT does not cover enterprise/. Cloud paid plans also lead with a 15-day trial. Hacker is live-chat-only. ## Verdict The MIT-core inbox. Self-host is the free forever path; Cloud Hacker is a cap, not a product. --- # Socket.IO > MIT realtime library with fallbacks, not a hosted product **Category:** Comms **Pricing:** open-source **URL:** https://socket.io **GitHub:** https://github.com/socketio/socket.io **Added:** 2026-09-03 --- ## The Hook Socket.IO is still the WebSocket library people actually ship when the connection has to survive a proxy. ## What's Good MIT. Engine.IO fallbacks (WebSocket, HTTP long-polling), rooms, acknowledgements, binary. Client and server that have been in production for a decade-plus. No product pricing page because there is no product — you run the server. ## The Catch You bring the host. Horizontal scale needs Redis (or equivalent) adapter. This is not LiveKit and not a managed pub-sub. CORS, sticky sessions, and polling surprises remain your bugs. ## Verdict The MIT socket library. Free as in you run it. --- # Plunk > AGPL-3.0 transactional email, workflows, and broadcasts **Category:** Comms **Pricing:** freemium **URL:** https://www.useplunk.com **GitHub:** https://github.com/useplunk/plunk **Added:** 2026-09-03 --- ## The Hook Plunk is Resend plus campaigns, if you can live with AGPL and a branded footer. ## What's Good AGPL-3.0. Transactional, workflow automation, and broadcasts in one. Hosted Free is 1,000 emails/month, no credit card. Self-host with Docker Compose. Unlimited contacts on the pricing page. ## The Catch AGPL-3.0 is copyleft. Hosted Free keeps Plunk branding. This is not Resend at volume — 1k/mo is the free wall, then $0.001/email. Self-host means you bring SES or equivalent. ## Verdict AGPL email with a real 1k hosted free tier. Branding stays until you pay. --- # Centrifugo > Apache-2.0 Go realtime server binary, not in-process Socket.IO **Category:** Comms **Pricing:** open-source **URL:** https://centrifugal.dev **GitHub:** https://github.com/centrifugal/centrifugo **Added:** 2026-09-03 --- ## The Hook Centrifugo is the Go realtime server you run next to your app, not a library inside it. ## What's Good Apache-2.0. Language-agnostic PUB/SUB: WebSocket, HTTP-streaming, SSE, gRPC, WebTransport. Official SDKs. docker run centrifugo. Decouples transport from business logic. ## The Catch You run it. This is not Socket.IO in-process. Cloud/PRO SKUs were unverified here — no invented Cloud prices. Ops, clustering, and Redis when you scale stay yours. ## Verdict Apache-2.0 realtime binary. Free as in you operate the server. --- # Liveblocks > Hosted collab rooms; Free $0 pauses at the cap **Category:** Comms **Pricing:** freemium **URL:** https://liveblocks.io **GitHub:** https://github.com/liveblocks/liveblocks **Added:** 2026-09-03 --- ## The Hook Liveblocks is the hosted collab layer you rent when Socket.IO is the wrong job. ## What's Good Free is $0, no credit card. Caps: 3,000 collab minutes, 200 comments, 1 GB realtime storage, 512 MB files, 10 connections/room, 3 dashboard seats. Watermark required. Presence, comments, Yjs, storage. Pair with /skills/liveblocks-best-practices for the cookbook — this page is the product Catch, not the skill. ## The Catch Free pauses at the cap until the next cycle — not PAYG. Pro is $30/mo (credits + drop branding). Self-host is an Enterprise add-on, not the OSS path. Distinct from Socket.IO (you run the server) and Centrifugo (Go binary you operate). ## Verdict Hosted collab with a real $0 Free. Caps pause the feature; Pro is $30; self-host is sales. --- # Yjs > MIT CRDT library, not a hosted collab product **Category:** Comms **Pricing:** open-source **URL:** https://github.com/yjs/yjs **GitHub:** https://github.com/yjs/yjs **Added:** 2026-09-03 --- ## The Hook Yjs is the CRDT you import. There is no Yjs Cloud, no SKU, no watermark. ## What's Good MIT. Shared types for building collaborative software. Pair with Liveblocks if you want rooms hosted, or bring your own provider. Pair with /skills/yjs-best-practices for the library cookbook — this gem is the Catch, not a duplicate skill page. ## The Catch Not a product. No hosted SKU, no pricing page, no support SLA. You bring the network layer (Liveblocks, y-websocket, your own). Distinct from Socket.IO: CRDT merge, not event fan-out. ## Verdict The MIT CRDT. Pair with Liveblocks for hosted rooms; this is the library, not the meter. --- # Soketi > AGPL-3.0 Pusher Protocol v7 server you run yourself **Category:** Comms **Pricing:** open-source **URL:** https://docs.soketi.app **GitHub:** https://github.com/soketi/soketi **Added:** 2026-09-03 --- ## The Hook Soketi is the Pusher protocol without the Pusher invoice. ## What's Good AGPL-3.0. Implements Pusher Protocol v7 — existing Pusher clients plug in. Self-host the WebSocket server. App keys/secrets, channels, broadcasting. Cloudflare Workers path exists in the README. ## The Catch Drop-in Pusher-protocol server, not Socket.IO in-process and not Centrifugo language-agnostic PUB/SUB. AGPLv3 is network copyleft. Hosted plans were not verified here — none invented. ## Verdict AGPL Pusher-protocol server. Free as in you run it; do not assume a Cloud SKU. --- # Mercure > AGPL-3.0 SSE hub; homepage still says MIT **Category:** Comms **Pricing:** open-source **URL:** https://mercure.rocks **GitHub:** https://github.com/dunglas/mercure **Added:** 2026-09-03 --- ## The Hook Mercure is the SSE hub whose marketing page has not caught up with its LICENSE file. ## What's Good LICENSE file is GNU AGPLv3. SSE hub: publish with POST, subscribe with EventSource. Caddy-based, Docker/Helm, Last-Event-ID replay. Official realtime path for Symfony. Self-host the Hub. ## The Catch Homepage still says the Hub is MIT-licensed. The LICENSE file is AGPL-3.0 — print the file. SSE, not WebSockets, not Pusher protocol. Distinct from Socket.IO and Soketi. Cloud/Enterprise exist; do not invent prices. OSS self-host is the free path. ## Verdict AGPL SSE hub. Believe the LICENSE file, not the MIT line on the homepage. --- # NATS > Apache-2.0 subject-based messaging and streaming fabric **Category:** Comms **Pricing:** open-source **URL:** https://nats.io **GitHub:** https://github.com/nats-io/nats-server **Added:** 2026-09-04 --- ## The Hook NATS is the subject-based messaging fabric you run when sockets alone are the wrong layer. ## What's Good Apache-2.0. OSS self-host at $0. Pub/sub, request/reply, queue groups, JetStream streaming. Tiny Go server, clients across languages. CNCF project. ## The Catch Subject-based messaging/streaming fabric — not an in-process WS lib and not a Go realtime binary alone. Distinct from Socket.IO and Centrifugo. ## Verdict Apache-2.0 messaging fabric. Free as in you run the server. Not Socket.IO. --- # AnyCable > MIT realtime for Action Cable with delivery guarantees **Category:** Comms **Pricing:** freemium **URL:** https://anycable.io **GitHub:** https://github.com/anycable/anycable **Added:** 2026-09-04 --- ## The Hook AnyCable is Action Cable that survives scale without pretending to be Pusher. ## What's Good MIT. Open-source tier is Free. Delivery guarantees, presence, Action Cable protocol, multi-backend. Drop-in path for Rails apps that outgrew the default cable server. ## The Catch Delivery guarantees + presence + Action Cable / multi-backend — not a Pusher-protocol server. Distinct from Socket.IO and Centrifugo. Pro and Managed exist; prices not reprinted here. ## Verdict MIT Action Cable path. OSS tier is free; not Soketi, not Socket.IO. --- # mediasoup > ISC embeddable SFU as a Node module or Rust crate **Category:** Comms **Pricing:** open-source **URL:** https://mediasoup.org **GitHub:** https://github.com/versatica/mediasoup **Added:** 2026-09-04 --- ## The Hook mediasoup is the SFU you embed, not the rooms product you rent. ## What's Good ISC. Node package and Rust crate. You own signaling and the room model. ## The Catch Embeddable media router as Node module/Rust crate — not LiveKit rooms/agents product. Distinct from LiveKit. ## Verdict ISC embeddable media router. Free as in you wire it. Not LiveKit. --- # PeerJS > MIT browser P2P WebRTC with free cloud or self-host signaling **Category:** Comms **Pricing:** freemium **URL:** https://peerjs.com **GitHub:** https://github.com/peers/peerjs **Added:** 2026-09-04 --- ## The Hook PeerJS is browser-to-browser WebRTC when you do not want to stand up a media-server room product. ## What's Good MIT. Data and media connections in the browser. Free cloud PeerServer or self-host your own signaling. Tiny API for 1:1 and small meshes. ## The Catch Browser P2P WebRTC + optional free signaling — not LiveKit/mediasoup media-server rooms. Larger groups need a media server (SFU-class), not PeerJS alone. Distinct from LiveKit and mediasoup. ## Verdict MIT P2P WebRTC. Free signaling path; not a rooms SFU. --- # Mautic > GPL-3.0 self-host marketing automation; hosted trial is not free **Category:** Comms **Pricing:** open-source **URL:** https://www.mautic.org **GitHub:** https://github.com/mautic/mautic **Added:** 2026-09-04 --- ## The Hook Mautic is marketing automation you self-host when Listmonk is only newsletters. ## What's Good GPL-3.0. Campaigns, segments, scoring, landing pages, forms — the automation suite. Self-host is the free path. Community and agency ecosystem. ## The Catch Marketing automation — not Listmonk newsletter/tx alone. Hosted 14-day trial is NOT the free path; self-host is. Distinct from Listmonk. ## Verdict GPL-3.0 marketing automation. Self-host free; hosted trial is not forever-$0. --- # Janus Gateway > GPL-3.0 plugin-core WebRTC gateway (SIP, VideoRoom, streaming) **Category:** Comms **Pricing:** open-source **URL:** https://janus.conf.meetecho.com **GitHub:** https://github.com/meetecho/janus-gateway **Added:** 2026-09-04 --- ## The Hook Janus is the plugin-core C gateway when you need SIP, VideoRoom, streaming, and recording — not an opinionated rooms SaaS. ## What's Good GPL-3.0 core. Plugin architecture: SIP, VideoRoom, streaming, recording, and more. Battle-tested Meetecho gateway. Commercial options exist from Meetecho — do not invent prices; OSS self-host is the free path. ## The Catch Plugin-core C gateway — not LiveKit/mediasoup opinionated rooms products. You pick plugins and ops the gateway. Distinct from LiveKit, mediasoup, PeerJS, and Pion. ## Verdict GPL-3.0 WebRTC gateway. Self-host free; commercial options exist — pitch the OSS core. --- # SocketCluster > MIT Node pub/sub+RPC+JWT with SCC scale-out **Category:** Comms **Pricing:** open-source **URL:** https://socketcluster.io **GitHub:** https://github.com/SocketCluster/socketcluster **Added:** 2026-09-04 --- ## The Hook SocketCluster is Node realtime with pub/sub, RPC, and JWT when Socket.IO is only sockets. ## What's Good MIT. Pub/sub + RPC + JWT auth in one Node stack. SCC (SocketCluster Cluster) for horizontal scale-out. You run the servers — no product seat tax on the core. ## The Catch Node pub/sub+RPC+JWT+SCC scale-out — not Socket.IO library-only, not Soketi Pusher-protocol, not Centrifugo Go hub, not NATS messaging, not AnyCable Rails WS. Distinct from Socket.IO, Soketi, Centrifugo, NATS, and AnyCable. ## Verdict MIT Node realtime cluster. Self-host free; not Socket.IO alone. --- # Laravel Reverb > MIT first-party Laravel WebSocket server **Category:** Comms **Pricing:** open-source **URL:** https://laravel.com/docs/reverb **GitHub:** https://github.com/laravel/reverb **Added:** 2026-09-04 --- ## The Hook Laravel Reverb is first-party WebSockets when Soketi is a Pusher clone and you already live in Laravel. ## What's Good MIT. First-party Laravel WebSocket server. Echo-compatible. No third-party Pusher protocol tax. You run it next to your app. ## The Catch First-party Laravel WS — not Soketi Pusher-protocol, not Centrifugo Go hub, not AnyCable Rails, not Socket.IO library, not Mercure SSE. Distinct from Soketi, Centrifugo, AnyCable, Socket.IO, and Mercure. ## Verdict MIT Laravel WebSockets. Free with your Laravel stack. --- # Tuwunel > Apache-2.0 Matrix homeserver; official conduwuit successor **Category:** Comms **Pricing:** open-source **URL:** https://github.com/matrix-construct/tuwunel **GitHub:** https://github.com/matrix-construct/tuwunel **Added:** 2026-09-04 --- ## The Hook Tuwunel is the maintained Matrix homeserver after conduwuit was archived. ## What's Good Apache-2.0. Official successor to archived conduwuit. Maintained Matrix homeserver you self-host. Federation-capable without Synapse Java weight. ## The Catch Official conduwuit successor — Conduit is still beta; do not dual-KEEP Conduit. Matrix homeserver slot, not a generic WS product. Distinct from LiveKit/Soketi-style realtime. ## Verdict Apache-2.0 Matrix homeserver. Maintained successor; skip beta Conduit dual-keep. --- # Eclipse Mosquitto > EPL-2.0 / BSD-3 MQTT broker; OSS self-host forever free **Category:** Comms **Pricing:** open-source **URL:** https://mosquitto.org **GitHub:** https://github.com/eclipse-mosquitto/mosquitto **Added:** 2026-09-04 --- ## The Hook Mosquitto is the MQTT broker you run when NATS subjects are the wrong protocol. ## What's Good EPL-2.0 OR BSD-3-Clause (EDL-1.0; LICENSE.txt SPDX). OSS self-host forever free. Lightweight MQTT broker. Cedalo Pro is optional commercial — not required for the free path. ## The Catch MQTT broker, not a NATS messaging fabric. Distinct from Soketi/Laravel Reverb/Centrifugo (WebSocket/Pusher paths). Not a messaging-bus twin of NATS. Cedalo Pro exists; do not invent rates. ## Verdict Eclipse MQTT broker. Self-host free forever. Not NATS, not Soketi. --- # Sockudo > MIT Rust Pusher V1 edge + Protocol V2 recovery/history/push **Category:** Comms **Pricing:** open-source **URL:** https://sockudo.io **GitHub:** https://github.com/sockudo/sockudo **Added:** 2026-09-04 --- ## The Hook Sockudo is the Rust realtime server that keeps Pusher clients and adds durable Protocol V2 when you outgrow fire-and-forget. ## What's Good MIT. Self-host Rust + Tokio. Protocol V1 keeps pusher-js / Laravel Echo. Protocol V2 adds recovery, history, mutable messages, push, AI Transport. Horizontal fanout adapters. ## The Catch Vs Soketi: Rust Pusher V1 edge + Protocol V2 recovery/history/mutable/push — not AGPL v7-only. Distinct from Soketi AGPL Protocol v7 drop-in. Self-host is the free path; do not invent Cloud SKUs. ## Verdict MIT Rust Pusher-compat with V2 durability. Not AGPL Soketi. --- # PartyKit > MIT realtime rooms on Cloudflare Durable Objects; pay CF Workers only **Category:** Comms **Pricing:** open-source **URL:** https://www.partykit.io **GitHub:** https://github.com/cloudflare/partykit **Added:** 2026-09-04 --- ## The Hook PartyKit is stateful multiplayer rooms as code when Liveblocks is the hosted collab suite and raw WebSockets are the weekend you lose. ## What's Good MIT (Cloudflare-owned). PartyServer model on Workers + Durable Objects. TypeScript rooms, Yjs-friendly patterns, partykit deploy to your CF account. No PartyKit seat tax — Workers free tier covers small traffic. ## The Catch Vs Liveblocks/Yjs+Hocuspocus: room primitive on CF, not prebuilt comments/inbox SaaS. You pay Cloudflare Workers/DO usage at scale. Auth, persistence, and CRDT choice are yours. Post-acquisition roadmap rides CF. ## Verdict MIT CF realtime rooms. Free software; hosting is Workers/DO billing. --- # ⚑ Notifications Push, email, and pub-sub that fire when something actually happened. *14 gems in this category* --- # ntfy > HTTP pub-sub notifications you can curl from cron **Category:** Notifications **Pricing:** open-source **URL:** https://ntfy.sh **GitHub:** https://github.com/binwiederhier/ntfy **Added:** 2026-09-03 --- ## The Hook ntfy is push notifications without an account, a SDK, or a sales call. ## What's Good Dual-licensed Apache-2.0 OR GPL-2.0. curl -d to a topic and phones ring. First-party CLI, Android/iOS apps, single-binary server. Public ntfy.sh is free without signup. Self-host if you want reserved topics and no shared rate limits. ## The Catch Anonymous topics are passwords — guessable names leak other people's alerts. Public ntfy.sh rate-limits; reserved topics need Pro. Dual license means you pick Apache-2.0 or GPLv2, not Apache-only. ## Verdict The notification CLI that is actually free. curl it from a cron job. --- # Novu > MIT notification workflows with a real 0 USD Cloud **Category:** Notifications **Pricing:** freemium **URL:** https://novu.co **GitHub:** https://github.com/novuhq/novu **Added:** 2026-09-03 --- ## The Hook Novu is notification infrastructure that publishes MIT — and a default branch with no LICENSE file. ## What's Good Multi-channel workflows (email, in-app, SMS, chat, push), Inbox component, digest/delay. Cloud Free is 0 USD/month, no credit card: 10,000 workflow runs, all channels, US+EU residency. Pro from 30 USD/month. LICENSE-MIT exists on next; main has a LICENSE file. ## The Catch The default GitHub branch is next, and GET /LICENSE there 404s. Open-core: enterprise/, apps/web/src/ee, and apps/dashboard/src/ee are commercial. Self-hosted deployment on the pricing page sits under Enterprise. Free Cloud caps 10K runs, 20 workflows, 24h retention. ## Verdict MIT notification layer with a 0 USD Cloud. Default-branch LICENSE 404 is the catch. --- # Listmonk > AGPL-3.0 self-hosted newsletters and transactional mail **Category:** Notifications **Pricing:** open-source **URL:** https://listmonk.app **GitHub:** https://github.com/knadh/listmonk **Added:** 2026-09-03 --- ## The Hook Listmonk is Mailchimp if Mailchimp were a single binary and a Postgres database. ## What's Good AGPL-3.0. Campaigns, lists, templates, and transactional API. You run it; there is no vendor meter. Fast Go binary, first-party docs, no seat tax. ## The Catch AGPL-3.0 is copyleft — network use means you share modifications. You operate Postgres, SMTP, and bounces. There is no official hosted Listmonk Cloud. ## Verdict The self-host newsletter stack. AGPL, fully free, you run the mail. --- # Apprise > BSD-2-Clause CLI that talks to a hundred notifiers **Category:** Notifications **Pricing:** open-source **URL:** https://github.com/caronc/apprise **GitHub:** https://github.com/caronc/apprise **Added:** 2026-09-03 --- ## The Hook Apprise is one CLI, many notifiers, zero product page. ## What's Good BSD-2-Clause. apprise -t Title -b Body and it fans out to Slack, Discord, email, SMS, and a hundred more. Python library plus CLI. You point URLs at services you already have. ## The Catch Not a hosted inbox. There is no pricing page — that is the path. You bring every backend (SMTP, Slack tokens, Twilio). This is a notifier, not ntfy and not Novu. ## Verdict The notification Swiss Army CLI. BSD, no meter, no product SKU. --- # Gotify > MIT self-host push server; Android-first, no ntfy.sh twin **Category:** Notifications **Pricing:** open-source **URL:** https://gotify.net **GitHub:** https://github.com/gotify/server **Added:** 2026-09-03 --- ## The Hook Gotify is ntfy without the public hosted twin — you run the server or you do not get push. ## What's Good MIT. Self-host server (REST + WebSocket + WebUI), Android client, first-party CLI. Send via API; subscribe via websocket. You own the box. ## The Catch No public hosted twin like ntfy.sh. Android-first client, not a public pub-sub host. Distinct from ntfy (public ntfy.sh + dual license) and Apprise (CLI fan-out, no inbox). ## Verdict MIT push you self-host. No ntfy.sh-shaped public instance; Android is the client. --- # Edgepush > AGPL server + MIT SDK: BYO APNs/FCM on Cloudflare Workers **Category:** Notifications **Pricing:** freemium **URL:** https://edgepush.dev/pricing **GitHub:** https://github.com/akshitkrnagpal/edgepush **Added:** 2026-09-04 --- ## The Hook Edgepush is device push on Cloudflare Workers when you bring your own APNs and FCM keys. ## What's Good Server AGPL-3.0 self-host is $0. SDK/CLI MIT. Hosted Free: 1 app / 10k events/mo. BYO APNs+FCM — no middleman push vendor lock on credentials. Do not reprint Pro pricing. ## The Catch BYO APNs+FCM on Cloudflare Workers — not ntfy/Gotify self-host pub/sub, not Novu workflow orchestration. Server is AGPL; pitch self-host or Hosted Free caps. Distinct from ntfy, Gotify, and Novu. ## Verdict AGPL Workers push. Self-host $0; Hosted Free is capped — ignore Pro. --- # Shove > MIT end-to-end Go device-push relay **Category:** Notifications **Pricing:** open-source **URL:** https://github.com/pennersr/shove **GitHub:** https://github.com/pennersr/shove **Added:** 2026-09-04 --- ## The Hook Shove is MIT Go device-push when Edgepush is AGPL on Workers. ## What's Good MIT end-to-end. Go relay for device push. Self-host with no AGPL server clause. Simple ops path for APNs/FCM-style delivery without a workflow product. ## The Catch Edgepush server is AGPL; Shove is MIT end-to-end Go. Device-push relay — not ntfy/Gotify general pub/sub, not Novu multi-channel workflows. Distinct from Edgepush, ntfy, Gotify, and Novu. ## Verdict MIT Go push relay. Self-host free; not Edgepush AGPL Workers. --- # Postal > MIT self-hosted mail delivery: SMTP + HTTP API **Category:** Notifications **Pricing:** open-source **URL:** https://github.com/postalserver/postal **GitHub:** https://github.com/postalserver/postal **Added:** 2026-09-04 --- ## The Hook Postal is the self-hosted mail server when Resend is hosted API and Listmonk is newsletters. ## What's Good MIT. SMTP + HTTP API mail delivery you run. Full control of queues, IPs, and logs. Docs live via GitHub (postalserver.io 301s — use the repo). No per-seat tax on the core. ## The Catch Self-hosted mail delivery SMTP+HTTP API — not Resend hosted tx email, not Plunk Node email, not Listmonk newsletter app. You own deliverability and IP reputation. Distinct from Resend, Plunk, and Listmonk. ## Verdict MIT self-host mail. Free to run; deliverability is your ops. --- # Hyvor Relay > AGPL-3.0 email relay; Self-Hosted Free Forever unlimited **Category:** Notifications **Pricing:** open-source **URL:** https://relay.hyvor.com **GitHub:** https://github.com/hyvor/relay **Added:** 2026-09-04 --- ## The Hook Hyvor Relay is AGPL email relay you self-host forever when Postal is a full mail stack. ## What's Good AGPL-3.0. Self-Hosted Free Forever unlimited — that is the free path. Relay-focused email sending without claiming Cloud free. Pitch self-host only. ## The Catch Do NOT claim Cloud free. Self-host unlimited is the pitch. Vs Postal full mail server and Plunk Node email. Distinct from Postal and Plunk. ## Verdict AGPL email relay. Self-host free forever; no Cloud free claim. --- # Textbee > MIT Android SMS gateway; Hosted Free 1 device/50 msgs/day **Category:** Notifications **Pricing:** freemium **URL:** https://textbee.dev **GitHub:** https://github.com/vernu/textbee **Added:** 2026-09-04 --- ## The Hook Textbee is SMS through your Android phone when Twilio trial credits are the wrong free path. ## What's Good MIT. Hosted Free: 1 device / 50 msgs/day / 300/mo. Self-host the gateway. Turn a spare Android into an SMS API without buying a CPaaS seat. ## The Catch Android phone must stay online; not iOS. SMS hole vs Twilio trials — you own the handset and carrier, not a cloud number product. Distinct from Twilio-class CPaaS. ## Verdict MIT Android SMS gateway. Hosted Free is capped; keep the phone online. --- # Mailu > MIT full mailbox stack: SMTP/IMAP/webmail — not tx API **Category:** Notifications **Pricing:** open-source **URL:** https://mailu.io **GitHub:** https://github.com/Mailu/Mailu **Added:** 2026-09-04 --- ## The Hook Mailu is full email hosting you self-host when Postal is outbound delivery and Resend is transactional API. ## What's Good MIT. SMTP + IMAP + webmail in one Docker-friendly stack. Real mailboxes for users, not just an HTTP send endpoint. Self-host is free. ## The Catch Full mailbox SMTP/IMAP/webmail — not Resend/Plunk/Postal transactional-or-relay API. Ops-heavy: DNS, TLS, spam, storage. Distinct from Postal, Plunk, and Resend. ## Verdict MIT mailbox suite. Free to run; ops-heavy vs a tx email API. --- # Knock > Hosted notification workflows; Developer free to 10k messages/mo **Category:** Notifications **Pricing:** freemium **URL:** https://knock.app **GitHub:** https://github.com/knocklabs **Added:** 2026-09-04 --- ## The Hook Knock is the managed cross-channel notification fabric when Novu is the self-host twin and you refuse to own the inbox plumbing. ## What's Good Workflows + preferences + in-app feeds/guides across email, push, SMS, Slack, Teams, Discord. Developer $0: 10k messages/mo, unlimited workflows/channels/members, batch/delay, multi-tenancy, Google SSO. SDKs + CLI. Charge only for messages sent (not failed/bounced triggers). ## The Catch Closed-source SaaS — not Novu MIT self-host. LOUD: Starter jumps to $250/mo (50k messages). Downstream SendGrid/Twilio/etc still bill you. HIPAA/SAML/SCIM are Enterprise. Free path is Developer tier only. ## Verdict Hosted notification workflows. Free to 10k msgs; Starter $250. Not self-hostable. --- # OneSignal > Multi-channel messaging SaaS; Free now capped at 1k mobile MAU **Category:** Notifications **Pricing:** freemium **URL:** https://onesignal.com **Added:** 2026-09-04 --- ## The Hook OneSignal is the push/email/in-app suite when Knock is workflow-orchestration-first and ntfy is self-host curl. ## What's Good Push (mobile+web), in-app, email, Journeys. Free still useful for small apps: web push (10k/send), 10k emails/mo, mobile push/in-app under the Free MAU cap. Mature SDKs and dashboard. ## The Catch Closed SaaS. LOUD (2026): Free mobile push/in-app limited to 1,000 MAU (new orgs Sep 1, existing Oct 1) — overage pauses mobile until Growth (from about $19/mo + per-MAU). Not Novu/Knock twins. SMS/RCS extra. ## Verdict Messaging SaaS. Free for tiny MAU; Growth meters mobile hard after 1k. --- # MagicBell > Multichannel notification API + inbox; Builder free to 1k deliveries/mo **Category:** Notifications **Pricing:** freemium **URL:** https://www.magicbell.com **Added:** 2026-09-04 --- ## The Hook MagicBell is the notification inbox + routing fabric when Knock is workflow-orchestration-first and OneSignal is push-marketing-heavy. ## What's Good One API across in-app, email, mobile/web push, Slack, SMS. Built-in preferences, smart delivery, delivery observability. Builder $0: 1k deliveries/mo, all integrations, unlimited users/members, 1 project. ## The Catch Closed SaaS — not Novu self-host. LOUD: Startup jumps to $249/mo (50k deliveries; $0.0025 overage). Multi-channel = multi-delivery counts. You still pay SendGrid/Twilio/FCM underneath. Not Knock twin for complex workflow DAGs. ## Verdict Notification inbox SaaS. Free to 1k deliveries; Startup $249. --- # ▤ Libraries Frameworks and engines you import, not dashboards you rent. *38 gems in this category* --- # Hono > Tiny TypeScript web framework that runs on every runtime **Category:** Libraries **Pricing:** open-source **URL:** https://hono.dev **GitHub:** https://github.com/honojs/hono **Added:** 2026-09-03 --- ## The Hook Hono is Express if Express had been designed for Workers, Deno, Bun, and Node at once. ## What's Good MIT. One API across Cloudflare Workers, Deno, Bun, Node, and AWS Lambda. Middleware, JSX, validation, and RPC helpers without a framework tax. The docs are short on purpose. If you are writing an API, not a full Next.js app, this is the default. ## The Catch Not Next.js. No filesystem router, no RSC, no batteries. You bring auth, ORM, and deploy. The Cloudflare-first energy is real — Node works, but the happy path is an edge runtime. ## Verdict The TypeScript HTTP framework when Next is the wrong shape. --- # Phaser > MIT HTML5 2D game framework with scenes, physics, and tilemaps **Category:** Libraries **Pricing:** open-source **URL:** https://phaser.io **GitHub:** https://github.com/phaserjs/phaser **Added:** 2026-09-03 --- ## The Hook Phaser is still the 2D browser-game framework a builder should start with. ## What's Good MIT (LICENSE.md). Phaser 4.2.1 Giedi. Scenes, Arcade + Matter physics, input, tilemaps, particles, cameras. Install the phaser package or pull jsDelivr. The framework stays OSS; named titles on phaser.io/games include Geometry Dash web demo and Stickforge. ## The Catch Phaser 4 replaced the v3 renderer/FX/mask stack — v3 games need the migration guide. Phaser Editor is a separate commercial desktop product, not the OSS engine. The homepage can Cloudflare-gate a raw fetch; docs and GitHub are the source of truth. ## Verdict Default 2D HTML5 game framework. MIT engine; Editor is paid and optional. --- # PixiJS > MIT 2D WebGPU/WebGL renderer, not a game engine **Category:** Libraries **Pricing:** open-source **URL:** https://pixijs.com **GitHub:** https://github.com/pixijs/pixijs **Added:** 2026-09-03 --- ## The Hook PixiJS is the 2D renderer studios ship when they already have a game engine in their head. ## What's Good MIT. Scene graph, assets, interaction, filters, WebGPU/WebGL. pixi.js 8.x. Showcase names Subway Surfers, LEGO City Adventures, Cartoon Network GameBox. No hosted pricing page — Open Collective is optional. ## The Catch A renderer, not a game engine. Bring physics, audio, scenes, netcode. If you wanted Phaser, this is the wrong layer. ## Verdict The 2D WebGPU/WebGL renderer. Not Phaser. Not a product. --- # Babylon.js > Apache-2.0 3D engine for the web, now on v9 **Category:** Libraries **Pricing:** open-source **URL:** https://www.babylonjs.com **GitHub:** https://github.com/BabylonJS/Babylon.js **Added:** 2026-09-03 --- ## The Hook Babylon.js is the production-grade open 3D engine Microsoft, Mojang, and Nike already shipped. ## What's Good Apache-2.0. Full 3D engine: scenes, physics, loaders, playground. Homepage is Babylon.js 9. Named experiences: Minecraft Classic, Nike By You, Temple Run 2, Xbox Design Lab. Registry package @babylonjs/core. ## The Catch Heavier than a 2D starter. docs.babylonjs.com (with an s) has returned 500; working docs host is doc.babylonjs.com. Editor pricing was not verified here, so it is omitted. ## Verdict The open 3D game and rendering engine for the web. Apache-2.0, on v9. --- # PlayCanvas > MIT WebGL2/WebGPU engine plus an optional hosted Editor **Category:** Libraries **Pricing:** freemium **URL:** https://playcanvas.com **GitHub:** https://github.com/playcanvas/engine **Added:** 2026-09-03 --- ## The Hook PlayCanvas is an OSS engine with a hosted Editor that charges for private projects. ## What's Good Engine MIT (Copyright 2011-2026 PlayCanvas Ltd.). WebGL2 + WebGPU, Gaussian splats. Registry package playcanvas, or self-host the engine. Hosted Editor Free is 0 USD/month: unlimited public projects, 1GB, download for self-hosting. README names King, Miniclip, Snap, Disney. ## The Catch Do not conflate the MIT engine with the hosted Editor. Private cloud projects are the paid line (Personal 15 USD/month; Organization 50 USD per seat/month). Cancel and private projects lock until you resubscribe or make them public. ## Verdict OSS engine, optional Editor. Public projects are free; private cloud is not. --- # Colyseus > MIT Node multiplayer rooms with state sync **Category:** Libraries **Pricing:** open-source **URL:** https://colyseus.io **GitHub:** https://github.com/colyseus/colyseus **Added:** 2026-09-03 --- ## The Hook Colyseus is the authoritative multiplayer server a browser game can self-host for nothing. ## What's Good MIT. State sync, matchmaking, SDKs for JavaScript, Unity, Defold, Construct 3, Haxe. create-colyseus-app. Site: free forever, even for commercial games. Pair it with Phaser or KAPLAY. ## The Catch A server framework, not a renderer. Colyseus Cloud has no free tier — it starts at 15 USD/mo. Capacity depends on your code, not a marketed CCU cap. ## Verdict MIT multiplayer rooms. Self-host is the free path; Cloud is paid from dollar one. --- # KAPLAY > MIT successor to Kaboom for playground-first 2D games **Category:** Libraries **Pricing:** open-source **URL:** https://kaplayjs.com **GitHub:** https://github.com/kaplayjs/kaplay **Added:** 2026-09-03 --- ## The Hook KAPLAY is Kaboom, still alive, still MIT, still the 2D library when Phaser is too much surface. ## What's Good MIT. JS/TS 2D, component add() API, KAPLAYGROUND at play.kaplayjs.com. Site: open source, free and always will be. v3001 is the Kaboom-compatible line; v4000 is the future. Docs include a first-party Colyseus multiplayer guide. ## The Catch Two lines: stable v3001 vs v4000 still alpha. Official host is kaplayjs.com, not kaboomjs.com. Named homepage titles are testimonials, not a studio showcase. ## Verdict The living Kaboom successor. MIT, playground-first, not Phaser. --- # oRPC > MIT end-to-end typed TypeScript RPC **Category:** Libraries **Pricing:** open-source **URL:** https://orpc.dev **GitHub:** https://github.com/unnoq/orpc **Added:** 2026-09-03 --- ## The Hook oRPC is typed RPC that does not pretend to be tRPC with extra steps. ## What's Good MIT. End-to-end types, OpenAPI, standard schema, works across runtimes. Docs at orpc.dev. This is a library you import, not a hosted RPC product. ## The Catch TypeScript-first. Not a drop-in for every tRPC plugin. You still choose the HTTP server (Hono, Next, etc.). Do not confuse it with a backend-as-a-service. ## Verdict The typed TS RPC library. MIT, no meter, you bring the server. --- # howler.js > MIT audio sprites for the browser, tagged 2.2.4 in 2023 **Category:** Libraries **Pricing:** open-source **URL:** https://howlerjs.com **GitHub:** https://github.com/goldfire/howler.js **Added:** 2026-09-03 --- ## The Hook howler.js is still the audio-sprite library you drop next to Phaser or Pixi. ## What's Good MIT. Web Audio plus HTML5 fallbacks, sprites, spatial audio. ## The Catch Quiet since September 2023. Stale LICENSE copyright year -- live MIT line, not a ghost. No Used-By names. ## Verdict The browser audio sprite library. MIT, quiet, still the default next to Phaser. --- # Matter.js > MIT 2D rigid-body physics, not a Phaser replacement **Category:** Libraries **Pricing:** open-source **URL:** https://brm.io/matter-js **GitHub:** https://github.com/liabru/matter-js **Added:** 2026-09-03 --- ## The Hook Matter.js is the 2D physics engine Phaser already vendors — grab it when you are not on Phaser. ## What's Good MIT. 2D rigid bodies, constraints, collisions, sleeping. Drop next to PixiJS, KAPLAY, or a raw canvas. Docs and demos at brm.io/matter-js. ## The Catch Loud: Phaser already ships Matter physics. If you are on Phaser, you already have this. This gem is for Pixi/KAPLAY/canvas, not a second physics stack inside Phaser. Distinct from Phaser (full engine) and howler.js (audio sprites). ## Verdict The MIT 2D rigid-body lib. Use it off Phaser; on Phaser you already have Matter. --- # melonJS > MIT 2D engine: WebGPU to WebGL2 to Canvas AUTO, zero deps **Category:** Libraries **Pricing:** open-source **URL:** https://melonjs.org **GitHub:** https://github.com/melonjs/melonJS **Added:** 2026-09-04 --- ## The Hook melonJS is the Canvas2D-shaped 2D engine when Phaser plugin zoo is too much. ## What's Good MIT forever $0. Renderer cascade WebGPU then WebGL2 then Canvas AUTO. Canvas2D-shaped API. Zero dependencies. ## The Catch WebGPU to WebGL2 to Canvas AUTO and Canvas2D-shaped API — not Phaser scene/plugin ecosystem. W4 cut re-opened. Distinct from Phaser, KAPLAY, and PixiJS. ## Verdict MIT zero-dep 2D engine. Not Phaser. --- # Tone.js > MIT Web Audio: synths, Transport, effects, sample-accurate **Category:** Libraries **Pricing:** open-source **URL:** https://tonejs.github.io **GitHub:** https://github.com/Tonejs/Tone.js **Added:** 2026-09-04 --- ## The Hook Tone.js is musical Web Audio when howler.js is only sprites. ## What's Good MIT. Synths, Transport, effects, sample-accurate scheduling. Web Audio composition toolkit. ## The Catch Synths/Transport/effects/sample-accurate scheduling — not audio-sprite playback. W4 cut re-opened. Distinct from howler.js. ## Verdict MIT musical Web Audio. Not howler.js sprites. --- # Rapier > Apache-2.0 Rust/WASM physics: 2D+3D, joints, SIMD **Category:** Libraries **Pricing:** open-source **URL:** https://rapier.rs **GitHub:** https://github.com/dimforge/rapier **Added:** 2026-09-04 --- ## The Hook Rapier is the Rust/WASM physics engine when Matter.js stops at 2D MIT. ## What's Good Apache-2.0. 2D and 3D, joints, SIMD, optional determinism. Native Rust and WASM builds. Dimforge stack next to nalgebra. ## The Catch Rust/WASM 2D+3D with joints/SIMD/optional determinism — Matter.js is MIT 2D only. Planck HOLD (same 2D JS slot as Matter). Distinct from Matter.js. ## Verdict Apache-2.0 2D+3D physics. Not Matter.js 2D. --- # Pion WebRTC > MIT embeddable Go WebRTC library — not a rooms product **Category:** Libraries **Pricing:** open-source **URL:** https://pion.ly **GitHub:** https://github.com/pion/webrtc **Added:** 2026-09-04 --- ## The Hook Pion is pure-Go WebRTC you embed when LiveKit is a product and mediasoup is a Node media server. ## What's Good MIT. Pure Go WebRTC stack you import into your binary. Build custom peers, gateways, and media paths without adopting a rooms SaaS. First-class Go ecosystem. ## The Catch Embeddable Go library — not LiveKit rooms product, not mediasoup C++/Node media server, not PeerJS browser-only P2P helper. You assemble signaling and any SFU-class topology yourself. Distinct from LiveKit, mediasoup, and PeerJS. ## Verdict MIT Go WebRTC library. Free to embed; not a rooms stack. --- # Defold > Source-available Lua game engine; forever free, not OSI Open Source **Category:** Libraries **Pricing:** source-available **URL:** https://defold.com **GitHub:** https://github.com/defold/defold **Added:** 2026-09-04 --- ## The Hook Defold is the forever-free Lua engine for HTML5 and native when Phaser is JS-only canvas. ## What's Good Defold License v1.0. Forever free, no royalty. Lua scripting, HTML5 and native targets. Source available to read and build games. Small runtime, editor included. ## The Catch LOUD: source-available NOT OSI Open Source — §4a may not commercialize as a Game Engine Product. Forever free for making games; license is not MIT/Apache. Vs Phaser/melonJS/KAPLAY JS engines. Distinct from Phaser, melonJS, and KAPLAY. ## Verdict Forever-free Lua engine. Source-available — not OSI OSS; do not ship it as your own engine product. --- # Cute Framework > zlib/Unlicense C/C++ native-first framework + Emscripten **Category:** Libraries **Pricing:** open-source **URL:** https://randygaul.github.io/cute_framework/ **GitHub:** https://github.com/RandyGaul/cute_framework **Added:** 2026-09-04 --- ## The Hook Cute Framework is native-first C/C++ games when Phaser is JS canvas and Pixi is a renderer. ## What's Good zlib OR Unlicense. C/C++ native-first with Emscripten web path. Full framework (not just a renderer): windowing, input, audio, sprites. Permissive enough to ship without OSI drama. ## The Catch C/C++ native-first + Emscripten — not Phaser/KAPLAY JS engines. Full framework — not Pixi/Babylon/PlayCanvas renderer-or-editor slots alone. Distinct from Phaser, KAPLAY, Pixi, Babylon, and PlayCanvas. ## Verdict Permissive C/C++ game framework. Native-first; web via Emscripten. --- # Godot > MIT engine with HTML5/Wasm+WebGL2; Compatibility mode for embeds **Category:** Libraries **Pricing:** open-source **URL:** https://godotengine.org **GitHub:** https://github.com/godotengine/godot **Added:** 2026-09-04 --- ## The Hook Godot is the MIT full engine on the web when Phaser is a JS framework and you want an editor too. ## What's Good MIT. Full 2D/3D engine, editor, GDScript/C#. HTML5/Wasm + WebGL2 export. Forever free, no royalty. Huge asset and learning ecosystem. ## The Catch Loud: Compatibility/WebGL2 only for web; threaded export needs COOP/COEP — single-thread default since 4.3 for embeds. Not Phaser JS canvas. Distinct from Phaser. ## Verdict MIT Godot. Web works in Compatibility/WebGL2; plan COOP/COEP only if you need threads. --- # Heaps > MIT Haxe+HashLink engine; HTML5/WebGL; Dead Cells stack **Category:** Libraries **Pricing:** open-source **URL:** https://heaps.io **GitHub:** https://github.com/HeapsIO/heaps **Added:** 2026-09-04 --- ## The Hook Heaps is the Haxe game engine behind Dead Cells when Phaser is JS and Godot is GDScript. ## What's Good MIT. Haxe + HashLink with HTML5/WebGL targets. Proven on shipped titles (Dead Cells and friends). Cross-compile without a JS-only canvas story. ## The Catch Haxe+HashLink HTML5/WebGL — not Phaser/Godot/Defold/Cute Framework/melonJS/Bevy. Different language/runtime bet. Distinct from Phaser, Godot, Defold, Cute Framework, melonJS, and Bevy. ## Verdict MIT Haxe engine. HTML5/WebGL path; not a JS canvas framework. --- # raylib > zlib/libpng C game library + Emscripten web **Category:** Libraries **Pricing:** open-source **URL:** https://www.raylib.com **GitHub:** https://github.com/raysan5/raylib **Added:** 2026-09-04 --- ## The Hook raylib is simple C gamedev with an Emscripten web path when you refuse a JS engine. ## What's Good zlib/libpng. Batteries-included C API: windowing, input, audio, 2D/3D helpers. Emscripten builds for the browser. Tiny learning curve for native-first games. ## The Catch C+Emscripten web — not Phaser/Pixi/KAPLAY/melonJS JS engines. Lower-level than Godot/Defold/Bevy full engines. Distinct from JS canvas engines. ## Verdict zlib C game library. Web via Emscripten; not a JS framework. --- # GDevelop > MIT event/no-code engine + HTML5; Pro required above $50k revenue **Category:** Libraries **Pricing:** freemium **URL:** https://gdevelop.io **GitHub:** https://github.com/4ian/GDevelop **Added:** 2026-09-04 --- ## The Hook GDevelop is event-sheet games with HTML5 export when you want no-code scenes instead of Phaser code. ## What's Good MIT engine. Event/no-code + JS extensibility. HTML5 export. Free for most indie use via the account tiers. ## The Catch Loud: companies >$50k revenue must use Pro if using a GDevelop account. Event/no-code+HTML5 — not Phaser code-first or Godot editor. Distinct from Phaser and Godot. ## Verdict MIT no-code/HTML5 engine. Free until revenue/account rules force Pro. --- # matrix-js-sdk > Apache-2.0 Matrix client library — needs a homeserver **Category:** Libraries **Pricing:** open-source **URL:** https://github.com/matrix-org/matrix-js-sdk **GitHub:** https://github.com/matrix-org/matrix-js-sdk **Added:** 2026-09-04 --- ## The Hook matrix-js-sdk is the browser/Node Matrix client when Tuwunel is the homeserver you still have to run. ## What's Good Apache-2.0. Official Matrix client SDK for JS. Rooms, sync, E2EE paths in the client. Pair with any homeserver you choose. ## The Catch Client lib only — you need a homeserver (Tuwunel etc.). Not a homeserver and not a generic WS product. Distinct from Tuwunel. ## Verdict Apache-2.0 Matrix client. Free library; bring a homeserver. --- # RPG-JS > MIT browser RPG/MMORPG TypeScript framework; Free Engine $0 **Category:** Libraries **Pricing:** freemium **URL:** https://rpgjs.dev **GitHub:** https://github.com/RSamaium/RPG-JS **Added:** 2026-09-04 --- ## The Hook RPG-JS is the TypeScript RPG kit when Phaser alone is not enough MMORPG shape. ## What's Good MIT. Free Engine at $0. Same codebase for solo and networked play. Browser RPG/MMORPG framework in TypeScript. Studio is $99 lifetime optional — pitch the MIT engine / Free Engine only. ## The Catch Browser RPG/MMORPG TypeScript framework (same code solo + networked) — not Phaser scenes, not Colyseus rooms alone, not GDevelop no-code, not Godot/Heaps. Do not sell Studio as the free path. ## Verdict MIT Free Engine for browser RPGs. Studio is optional paid; engine is free. --- # Oxc > MIT Rust JS/TS toolchain: parser, linter, and friends **Category:** Libraries **Pricing:** open-source **URL:** https://oxc.rs **GitHub:** https://github.com/oxc-project/oxc **Added:** 2026-09-04 --- ## The Hook Oxc is the Rust JS/TS compiler substrate when you want parser/linter pieces, not a Biome replacement pitch. ## What's Good MIT. High-performance JS/TS parser, oxlint, transformer, minifier — building blocks for tooling. Written in Rust. Active project with a clear lower-level lane. ## The Catch Biome is live — Oxc is the lower-level toolchain / oxlint lane, not a Biome killer. Do not sell it as drop Prettier+ESLint tomorrow the way Biome markets format+lint as one product. ## Verdict MIT Rust JS/TS toolchain. Substrate and oxlint, not Biome. --- # Deno > MIT TypeScript runtime forever; Deploy free tier is real but separate **Category:** Libraries **Pricing:** freemium **URL:** https://deno.com **GitHub:** https://github.com/denoland/deno **Added:** 2026-09-04 --- ## The Hook Deno is the MIT TypeScript runtime you can keep when Node tooling tax is the complaint. ## What's Good MIT runtime forever. Secure-by-default permissions, built-in TypeScript, std library, npm interop matured. Deploy free tier is real. Skills already on the catalog; gem was missing. ## The Catch Deploy is not the runtime. The MIT runtime is free forever; Deno Deploy is a separate hosted product with a free tier — do not conflate them. Distinct from Node and Bun as runtimes. ## Verdict MIT runtime forever. Deploy free tier is real; do not merge the two. --- # Galacean Engine > MIT Ant Group TypeScript WebGL engine + browser Editor **Category:** Libraries **Pricing:** open-source **URL:** https://galacean.com/engine **GitHub:** https://github.com/galacean/engine **Added:** 2026-09-04 --- ## The Hook Galacean is the Ant Group TypeScript WebGL engine when Babylon or PlayCanvas is the wrong cultural fit. ## What's Good MIT. TypeScript WebGL engine with a browser Editor. Mobile-first interactive apps; WeChat minigame / Alipay mini-program export paths in the Editor story. Free to use per the site. ## The Catch Vs Babylon/PlayCanvas: Ant Group TS WebGL + browser Editor + WeChat minigame lane — not Microsoft Babylon docs gravity, not PlayCanvas hosted-Editor freemium. Do not invent Editor cloud prices here. ## Verdict MIT Ant Group WebGL engine. Editor + minigame path, not Babylon. --- # Excalibur.js > BSD-2 TypeScript 2D HTML5 game engine; free forever **Category:** Libraries **Pricing:** open-source **URL:** https://excaliburjs.com **GitHub:** https://github.com/excaliburjs/Excalibur **Added:** 2026-09-04 --- ## The Hook Excalibur is the batteries-included TypeScript 2D engine when Phaser is the bigger ecosystem and KAPLAY is the playground cousin. ## What's Good BSD-2-Clause. Scenes, actors, physics, tilemaps, loaders, TypeScript-first. Vite/Webpack templates. Commercial-friendly — no engine royalty. Active docs and samples. ## The Catch Still pre-1.0 API caution. Vs Phaser/melonJS/KAPLAY: smaller plugin marketplace than Phaser; not a 3D/WebGPU-first engine. You bring art pipeline and multiplayer. ## Verdict BSD-2 TS 2D engine. Free forever; pre-1.0, Phaser still wins ecosystem size. --- # Vite > MIT next-gen frontend tooling — dev server + build **Category:** Libraries **Pricing:** open-source **URL:** https://vite.dev **GitHub:** https://github.com/vitejs/vite **Added:** 2026-09-05 --- ## The Hook Vite is the default frontend toolchain when Next.js is the full framework and you still need a fast SPA/library build. ## What's Good MIT. Instant HMR, Rollup-based production builds, first-class TypeScript/JSX, huge plugin ecosystem. Powers Vue, Svelte, React SPAs, and library mode. The hole next to Next.js/Biome/Oxc in a modern JS stack. ## The Catch Bundler/dev-server — not an app framework (routing, SSR, data are yours or a meta-framework). Not Biome/Oxc (lint/format/parser). SSR frameworks built on Vite are separate products. ## Verdict MIT frontend toolchain. Free forever; framework and lint are adjacent gems. --- # Zod > MIT TypeScript-first schema validation with static type inference **Category:** Libraries **Pricing:** open-source **URL:** https://zod.dev **GitHub:** https://github.com/colinhacks/zod **Added:** 2026-09-05 --- ## The Hook Zod is the schema you write once that becomes both runtime checks and TypeScript types. ## What's Good MIT library. Parse unknown input to typed values. Composable objects, unions, transforms. De-facto standard across tRPC, forms, and API boundaries. Zero SaaS. ## The Catch TS schema plus static inference; library not SaaS. Bundle size matters at the edge. Not a form UI or API gateway. ## Verdict MIT TS schemas. Free library; you own the call sites. --- # Bun > MIT JS runtime + install + test + bundle (JSC linked) **Category:** Libraries **Pricing:** open-source **URL:** https://bun.sh **GitHub:** https://github.com/oven-sh/bun **Added:** 2026-09-05 --- ## The Hook Bun is the all-in-one JS toolkit when Node plus a separate installer/test/bundler stack is the tax you want to delete. ## What's Good MIT (JSC linked). Runtime, package install, test runner, and bundler in one binary. Node-compat target with speed as the pitch. Joined Anthropic; OSS stays MIT. ## The Catch Node-compat is not 100%. Joined Anthropic — watch ecosystem/governance noise, but license remains MIT. Not Deno. Some packages still assume Node quirks. ## Verdict MIT JS toolkit. Free forever; Node parity is the real catch. --- # TanStack Query > MIT server-state cache for async data (React Query lineage) **Category:** Libraries **Pricing:** open-source **URL:** https://tanstack.com/query **GitHub:** https://github.com/TanStack/query **Added:** 2026-09-06 --- ## The Hook TanStack Query is the server-state cache when fetching, caching, and invalidation should not live in your UI store. ## What's Good MIT. Declarative fetch/cache/invalidate for React, Vue, Solid, Svelte, Angular. Devtools, optimistic updates, infinite queries. De-facto async server-state default outside full RSC frameworks. ## The Catch Server-state cache library; not SaaS. Not a global client store (that is Zustand/Redux). Not tRPC itself — pairs with it. You still own the API and auth. ## Verdict MIT server-state cache. Free library; not a BaaS. --- # pnpm > MIT content-addressable Node pkg manager **Category:** Libraries **Pricing:** open-source **URL:** https://pnpm.io **GitHub:** https://github.com/pnpm/pnpm **Added:** 2026-09-06 --- ## The Hook pnpm shares a content-addressable store across monorepo installs. ## What's Good MIT. Strict layout, content-addressable store, workspaces. Disk-cheap vs classic npm. ## The Catch Vs Bun: Node pkg manager lane (content-addressable/monorepo); not install+runtime. Not a JS runtime or test runner. ## Verdict MIT Node pkg manager. Free forever; Bun is a different lane. --- # Vitest > MIT Vite-native unit/integration test runner **Category:** Libraries **Pricing:** open-source **URL:** https://vitest.dev **GitHub:** https://github.com/vitest-dev/vitest **Added:** 2026-09-06 --- ## The Hook Vitest is the test runner when your app already lives in Vite and Jest feels like a second bundler tax. ## What's Good MIT. Jest-compatible API, Vite transform pipeline, watch mode, TypeScript/ESM first-class, browser mode options. Shares config/plugins with Vite. Fast feedback in Vite apps and libraries. ## The Catch Vs Vite: Vite-native test runner; not the bundler. Not full E2E browser suites. Closest when Vite is already the toolchain. ## Verdict MIT Vite test runner. Free forever; Vite itself is the bundler gem. --- # Astro > MIT content-first web framework with islands **Category:** Libraries **Pricing:** open-source **URL:** https://astro.build **GitHub:** https://github.com/withastro/astro **Added:** 2026-09-06 --- ## The Hook Astro is the content-first framework when you want zero JS by default and islands for the interactive bits. ## What's Good MIT. Ships HTML-first pages, islands hydration, brings your own UI framework (React/Vue/Svelte/Solid). Great for docs, marketing, content sites. Content collections and adapters for SSR/hosting. ## The Catch Content-first islands; not Next/RSC. App-router SaaS dashboards often stay on Next. Server islands and adapters vary by host. Not a commerce-only stack. ## Verdict MIT islands framework. Free forever; Next is the RSC cousin. --- # Playwright > Apache-2.0 cross-browser library and CLI **Category:** Libraries **Pricing:** open-source **URL:** https://playwright.dev **GitHub:** https://github.com/microsoft/playwright **Added:** 2026-09-06 --- ## The Hook Playwright is the cross-browser library when tests and scripts need Chromium, Firefox, and WebKit from one API. ## What's Good Apache-2.0. Auto-waits, tracing, codegen, parallel projects. Official browsers and a strong CLI. Default for modern E2E in Node/.NET/Java/Python. ## The Catch Vs Playwright MCP gem: library/CLI lane; not the MCP wrapper. Heavier than unit tests. Browser downloads and CI images are real ops. Not Stagehand/Browserbase. ## Verdict Apache-2.0 browser library. Free; MCP wrapper is a separate gem. --- # SvelteKit > MIT full-stack Svelte framework **Category:** Libraries **Pricing:** open-source **URL:** https://svelte.dev/docs/kit **GitHub:** https://github.com/sveltejs/kit **Added:** 2026-09-06 --- ## The Hook SvelteKit is the full-stack Svelte lane when you want file-based routing, adapters, and SSR without React. ## What's Good MIT. Svelte compiler output, file routes, server load functions, form actions, adapters for Node/static/edge hosts. First-party story for Svelte apps beyond a SPA. ## The Catch Vs Remix/Hydrogen: full-stack Svelte lane. Ecosystem and hiring pool smaller than React/Next. Not Shopify Hydrogen commerce defaults. You still pick auth/data. ## Verdict MIT Svelte full-stack. Free forever; Remix/Hydrogen are React cousins. --- # fzf > MIT command-line fuzzy finder **Category:** Libraries **Pricing:** open-source **URL:** https://junegunn.github.io/fzf/ **GitHub:** https://github.com/junegunn/fzf **Added:** 2026-09-06 --- ## The Hook fzf is the fuzzy finder when shell history, files, and git refs should be filterable without leaving the keyboard. ## What's Good MIT. Blazing interactive filter for any list piped in. Shell keybindings, vim integration, preview windows. Universal CLI building block. ## The Catch Fuzzy finder only - not a file manager, not ripgrep itself (pairs with it), not a shell. You wire the pipes. Not Atuin shell history sync. ## Verdict MIT fuzzy finder. Free forever; you compose the workflows. --- # Elysia > MIT ergonomic Bun-first web framework **Category:** Libraries **Pricing:** open-source **URL:** https://elysiajs.com **GitHub:** https://github.com/elysiajs/elysia **Added:** 2026-09-06 --- ## The Hook Elysia is the Bun-native web framework when you want end-to-end types and OpenAPI without Express ceremony. ## What's Good MIT. Bun-first performance story, type-level validation (TypeBox/Standard Schema), plugins, OpenAPI generation. Ergonomic DX for APIs on Bun. ## The Catch Vs Hono: Bun-first ergonomic lane; Hono is multi-runtime edge-first. Ecosystem and hiring still Bun-skewed. Not Nest/Fastify full Node default. ## Verdict MIT Bun framework. Free forever; Hono is the multi-runtime cousin. --- # Turborepo > MIT Rust monorepo build system (JS/TS) **Category:** Libraries **Pricing:** open-source **URL:** https://turborepo.dev **GitHub:** https://github.com/vercel/turborepo **Added:** 2026-09-06 --- ## The Hook Turborepo is the monorepo task runner when JS and TS workspaces need a shared build graph. ## What's Good MIT. Fast task graph for JS/TS monorepos with caching and pruning. ## The Catch Task runner only. Not a bundler. Not Nx. Not Bazel. ## Verdict MIT monorepo builds. Free locally. --- # Skills 494 agent skills. Index: /skills # Claude Academy Guide > Claude Academy course matcher for product questions **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/academy-guide/SKILL.md --- ## When it is useful Stop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. ## The Catch Anthropic product skill. Recommends academy.claude.com courses — never invent them. Idle when the user wants the task done, not a tutorial. --- # Algorithmic Art > Seeded p5.js generative-art skill **Subcategory:** Frontend **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/algorithmic-art/SKILL.md --- ## When it is useful Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations. ## The Catch Idle unless someone asks for generative art. Original work only; not a copy-an-artist skill. --- # Anthropic Brand Guidelines > Anthropic brand colors and type for artifacts **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/brand-guidelines/SKILL.md --- ## When it is useful Applies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply. ## The Catch This is Anthropic's look — Poppins, Lora, #d97757 — not “your brand.” --- # Canvas Design > Poster and static visual design in PNG/PDF **Subcategory:** Frontend **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/canvas-design/SKILL.md --- ## When it is useful Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations. ## The Catch Static PNG/PDF posters. Not product UI. Will sit idle unless someone asks for a piece. --- # Claude API > Claude API / Anthropic SDK reference before you guess **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/claude-api/SKILL.md --- ## When it is useful - Reference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). ## The Catch Vendor + broad-trigger Catch: fires on almost any LLM-shaped task and steers to Anthropic. Skip when another provider is the job. --- # Doc Coauthoring > Structured loop for co-authoring specs and docs **Subcategory:** Authoring **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/doc-coauthoring/SKILL.md --- ## When it is useful Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. ## The Catch Anthropic’s three-stage writing loop. Never-fire if the agent just writes the doc. Distinct from Sentry’s copy in getsentry/skills. --- # Frontend Design > Intentional visual direction so UI is not template-default **Subcategory:** Frontend **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/frontend-design/SKILL.md --- ## When it is useful Guidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults. ## The Catch Aesthetic direction, not a component library. Idle unless someone is actually choosing a look. --- # MCP Builder > How to build MCP servers in Python or TypeScript **Subcategory:** MCP **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/mcp-builder/SKILL.md --- ## When it is useful Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK). ## The Catch How to *build* an MCP server. Not how to wire one that already exists. --- # Skill Creator > Create, edit, eval, and tune agent skills **Subcategory:** Authoring **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/skill-creator/SKILL.md --- ## When it is useful Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. ## The Catch Meta-skill for authoring and eval-ing skills. Idle unless you are writing a skill. --- # Slack GIF Creator > Animated GIFs sized and timed for Slack **Subcategory:** Other **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/slack-gif-creator/SKILL.md --- ## When it is useful Knowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack. ## The Catch Will sit idle unless someone actually asks for a Slack GIF. --- # Theme Factory > Ten preset themes, or a new one, for artifacts **Subcategory:** Frontend **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/theme-factory/SKILL.md --- ## When it is useful Toolkit for styling artifacts with a theme. These artifacts can be slides, docs, reportings, HTML landing pages, etc. There are 10 pre-set themes with colors/fonts that you can apply to any artifact that has been creating, or can generate a new theme on-the-fly. ## The Catch Ten palettes for artifacts. Idle unless someone wants a theme applied. --- # Web Artifacts Builder > Multi-component claude.ai HTML artifacts (React + shadcn) **Subcategory:** Frontend **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/web-artifacts-builder/SKILL.md --- ## When it is useful Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts. ## The Catch Locked to claude.ai HTML artifacts (React + Tailwind + shadcn bundle). Not a general web app skill. --- # Webapp Testing > Playwright toolkit for local webapp UI and logs **Subcategory:** Debugging **License:** Apache-2.0 **Source:** https://github.com/anthropics/skills/blob/main/skills/webapp-testing/SKILL.md --- ## When it is useful Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs. ## The Catch Playwright against localhost. Needs a running app; a screenshot is not a test plan. --- # Dispatching Parallel Agents > Split independent work across parallel agents **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/dispatching-parallel-agents/SKILL.md --- ## When it is useful Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies ## The Catch Only when 2+ tasks share no state. Sequential work is a false fire. --- # Executing Plans > Run a written plan in a separate session with checkpoints **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/executing-plans/SKILL.md --- ## When it is useful Use when you have a written implementation plan to execute in a separate session with review checkpoints ## The Catch Separate session, written plan, checkpoints. Not for writing the plan. --- # Finishing a Development Branch > Integrate a finished, tested branch **Subcategory:** Git **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/finishing-a-development-branch/SKILL.md --- ## When it is useful Use when implementation is complete, all tests pass, and you need to decide how to integrate the work ## The Catch Tests already pass. This decides how to land it — not whether it works. --- # Receiving Code Review > Verify review feedback before you implement it **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/receiving-code-review/SKILL.md --- ## When it is useful Use when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation ## The Catch Verify feedback before implementing. Rubber-stamping is the failure mode. --- # Requesting Code Review > Ask for review after a real feature or before merge **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/requesting-code-review/SKILL.md --- ## When it is useful Use when completing tasks, implementing major features, or before merging to verify work meets requirements ## The Catch After a real feature or before merge. Not a substitute for tests. --- # Subagent-Driven Development > Execute a plan with independent subagents in-session **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/subagent-driven-development/SKILL.md --- ## When it is useful Use when executing implementation plans with independent tasks in the current session ## The Catch In-session subagents. Conflicts with executing-plans (which wants a separate session). --- # Systematic Debugging > Root-cause the bug before proposing a fix **Subcategory:** Debugging **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/systematic-debugging/SKILL.md --- ## When it is useful Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes ## The Catch Evidence lives outside the agent. Reproduce the failure, read the log, then patch. Do not skip verification because the stack trace looked obvious. --- # Test-Driven Development > Write the failing test before implementation **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/test-driven-development/SKILL.md --- ## When it is useful Use when implementing any feature or bugfix, before writing implementation code ## The Catch Red, then green, then code. Agents skip the red step constantly. --- # Using Git Worktrees > Isolate feature work in a git worktree **Subcategory:** Git **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/using-git-worktrees/SKILL.md --- ## When it is useful Use when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback ## The Catch Isolate via worktree. No-op if you ignore it and stay on the dirty branch. --- # Verification Before Completion > Run the proof before claiming done **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/verification-before-completion/SKILL.md --- ## When it is useful Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always ## The Catch Run the proof before claiming done. Agents say “done” without running the command. --- # Writing Plans > Bite-sized implementation plan after a spec exists **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/obra/superpowers/blob/main/skills/writing-plans/SKILL.md --- ## When it is useful Use when you already have a spec or requirements for a multi-step task and need a bite-sized implementation plan before touching code — not for writing the spec itself. ## The Catch Bite-sized steps after a spec exists. Not for writing the spec (that’s spec-driven-development). --- # API and Interface Design > Stable APIs, module boundaries, and public contracts **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/api-and-interface-design/SKILL.md --- ## When it is useful Guides stable API and interface design. Use when designing APIs, module boundaries, or any public interface. Use when creating REST or GraphQL endpoints, defining type contracts between modules, or establishing boundaries between frontend and backend. ## The Catch Encyclopedia When from addyosmani/agent-skills. Never-fire if the agent does not load it. --- # Browser Testing with DevTools > Chrome DevTools MCP for real-browser evidence **Subcategory:** Debugging **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/browser-testing-with-devtools/SKILL.md --- ## When it is useful Tests in real browsers via Chrome DevTools MCP. Use when building or debugging anything that runs in a browser. Use when you need to inspect the DOM, capture console errors, analyze network requests, profile performance, or verify visual output with real runtime data. Requires the chrome-devtools MCP server to be configured. ## The Catch Requires the chrome-devtools MCP server. Without it, this is a prompt, not a browser. --- # CI/CD and Automation > Build, test, and deploy pipelines as quality gates **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/ci-cd-and-automation/SKILL.md --- ## When it is useful Automates CI/CD pipeline setup. Use when setting up or modifying build and deployment pipelines. Use when you need to automate quality gates, configure test runners in CI, or establish deployment strategies. ## The Catch Encyclopedia When from addyosmani/agent-skills. Not a CI vendor. Never-fire if the agent does not load it. --- # Code Review and Quality > Multi-axis review before a change lands **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/code-review-and-quality/SKILL.md --- ## When it is useful Conducts multi-axis code review. Use before merging any change. Use when reviewing code written by yourself, another agent, or a human. Use when you need to assess code quality across multiple dimensions before it enters the main branch. ## The Catch This is a review workflow, not AGENTS.md. Do not treat it as project agent instructions. --- # Code Simplification > Clarify working code without changing behavior **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/code-simplification/SKILL.md --- ## When it is useful Simplifies code for clarity. Use when refactoring code for clarity without changing behavior. Use when code works but is harder to read, maintain, or extend than it should be. Use when reviewing code that has accumulated unnecessary complexity. ## The Catch Clarify without changing behavior. Easy to “simplify” into a rewrite. --- # Constraint-Driven Development > Write the quality bar so agents cannot quietly lower it **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/constraint-driven-development/SKILL.md --- ## When it is useful Establishes a project's quality bar as a written contract and stops agents quietly lowering it. Interviews the user on which dimensions matter, supplies sane default thresholds when they have no number in mind, records everything in CONSTRAINTS.md, and watches the diff for a weakened bar — new @ts-ignore or eslint-disable suppressions, skipped or deleted tests, assertions stripped out, unimplemented stubs, thresholds edited down. Use when no quality bar is written down, when the user says "set up constraints" or "define our standards", when an agent keeps silencing checks or skipping tests to get to green, when you need a coverage or performance threshold and don't know what number to pick, or when an agent writes more code than anyone will read. ## The Catch Writes CONSTRAINTS.md. Useless if nobody reads it, or if the agent quietly lowers the bar anyway. --- # Deprecation and Migration > Sunset, migrate, and expand/contract without downtime **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/deprecation-and-migration/SKILL.md --- ## When it is useful Manages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code. ## The Catch Expand/contract, then delete. Not a license to drop a column on Friday. --- # Documentation and ADRs > Record decisions so the next engineer and agent can follow **Subcategory:** Authoring **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/documentation-and-adrs/SKILL.md --- ## When it is useful Records decisions and documentation. Use when making architectural decisions, changing public APIs, shipping features, or when you need to record context that future engineers and agents will need to understand the codebase. ## The Catch Record the decision. Agents write ADRs nobody asked for. --- # Doubt-Driven Development > Adversarial review of high-stakes decisions before they stand **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/doubt-driven-development/SKILL.md --- ## When it is useful Subjects every non-trivial decision to a fresh-context adversarial review before it stands. Use when correctness matters more than speed, when working in unfamiliar code, when stakes are high (production, security-sensitive logic, irreversible operations), or any time a confident output would be cheaper to verify now than to debug later. ## The Catch Adversarial review before the decision stands. Slow by design; skip it and you get confident wrongness. --- # Frontend UI Engineering > Production-quality accessible, responsive UI **Subcategory:** Frontend **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/frontend-ui-engineering/SKILL.md --- ## When it is useful Builds production-quality, accessible, responsive user-facing UIs. Use when building or modifying interfaces and pages, creating components, implementing layouts, meeting WCAG accessibility requirements, managing state, or when the output needs to look and feel production-quality rather than AI-generated. ## The Catch Production a11y/responsive UI. Overlaps frontend-design; this one is engineering, not vibe. --- # Git Workflow and Versioning > Commits, branches, PRs, tags, and version bumps **Subcategory:** Git **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/git-workflow-and-versioning/SKILL.md --- ## When it is useful Structures git workflow practices. Use when making any code change. Use when committing, branching, resolving conflicts, opening or reviewing a pull request (PR), pushing to a remote, or when you need to organize work across multiple parallel streams. Use when cutting a release, choosing a semantic version bump, tagging, or writing a changelog. ## The Catch Commits, PRs, tags. Not using-git-worktrees. Never-fire if the agent does not load it. --- # Idea Refine > Diverge, then converge a vague idea into a sharp concept **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/idea-refine/SKILL.md --- ## When it is useful Refines raw ideas into sharp, actionable concepts through structured divergent and convergent thinking. Use when an idea is still vague, when you need to stress-test assumptions before committing to a plan, or when you want to expand options before converging on one. Triggers on "ideate", "refine this idea", or "stress-test my plan". ## The Catch Diverge, then converge. Idle unless the idea is still vague. --- # Interview Me > One-question interviews until the real intent is clear **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/interview-me/SKILL.md --- ## When it is useful Extracts what the user actually wants instead of what they think they should want. Achieves this through one-question-at-a-time interview until ~95% confidence about the underlying intent. Use when an ask is underspecified ("build me X" without "for whom" or "why now"), when the user explicitly invokes ("interview me", "grill me", "are we sure?", "stress-test my thinking"), or when you catch yourself silently filling in ambiguous requirements before any plan, spec, or code exists. ## The Catch One question at a time until intent is clear. Annoying if you already know what you want. --- # Observability and Instrumentation > Logs, metrics, traces so production is diagnosable **Subcategory:** Debugging **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/observability-and-instrumentation/SKILL.md --- ## When it is useful Instruments code so production behavior is visible and diagnosable. Use when adding logging, metrics, tracing, or alerting. Use when shipping any feature that runs in production and you need evidence it works. Use when production issues are reported but you can't tell what happened from the available data. ## The Catch Logs, metrics, traces. Not Sentry-the-product. Never-fire if the agent does not load it. --- # Performance Optimization > Frontend, backend, and query performance with evidence **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/performance-optimization/SKILL.md --- ## When it is useful Optimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks. ## The Catch Needs a metric. Profiling theater without a number is not this skill. --- # Planning and Task Breakdown > Turn a spec into ordered, estimable tasks **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/planning-and-task-breakdown/SKILL.md --- ## When it is useful Breaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. ## The Catch After a spec. Overlaps writing-plans. Never-fire if the agent does not load it. --- # Security and Hardening > Input, auth, data, and supply-chain hardening **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/security-and-hardening/SKILL.md --- ## When it is useful Hardens code against vulnerabilities. Use when handling user input, authentication, data storage, or external integrations. Use when building any feature that accepts untrusted data, manages user sessions, or interacts with third-party services. Use when auditing dependencies for known vulnerabilities, triaging package-manager audit findings, or assessing supply-chain risk in a new package. Use when personal data or privacy compliance (GDPR, CCPA) is involved. ## The Catch Input, auth, data, supply-chain. Not a pentest and not Sentry security-review. --- # Shipping and Launch > Pre-launch checklist, rollout, and rollback **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/shipping-and-launch/SKILL.md --- ## When it is useful Prepares production launches. Use when preparing to deploy to production. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. ## The Catch Checklist, rollout, rollback. Not a deploy button. --- # Source-Driven Development > Cite official docs instead of remembered APIs **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/source-driven-development/SKILL.md --- ## When it is useful Grounds every implementation decision in official documentation. Use when you want authoritative, source-cited code free from outdated patterns. Use when building with any framework or library where correctness matters. ## The Catch Cite official docs. Agents still answer from memory. --- # Spec-Driven Development > Write the spec first when none exists yet **Subcategory:** Workflow **License:** MIT **Source:** https://github.com/addyosmani/agent-skills/blob/main/skills/spec-driven-development/SKILL.md --- ## When it is useful Use when starting a new project, feature, or significant change and no specification exists yet. Write the spec first. Distinct from Writing Plans, which assumes a spec already exists. ## The Catch Write the spec first when none exists. Distinct from writing-plans, which assumes a spec already exists. --- # Vercel React Best Practices > Vercel Engineering React and Next.js performance patterns **Subcategory:** Vercel **License:** MIT **Source:** https://github.com/vercel-labs/agent-skills/blob/main/skills/react-best-practices/SKILL.md --- ## When it is useful React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements. ## The Catch Vercel Engineering React/Next performance patterns. MIT in SKILL.md frontmatter; vercel-labs/agent-skills has no LICENSE file. Broad React triggers. --- # Vercel Composition Patterns > React composition that scales past boolean props **Subcategory:** Vercel **License:** MIT **Source:** https://github.com/vercel-labs/agent-skills/blob/main/skills/composition-patterns/SKILL.md --- ## When it is useful React composition patterns that scale. Use when refactoring components with boolean prop proliferation, building flexible component libraries, or designing reusable APIs. Triggers on tasks involving compound components, render props, context providers, or component architecture. Includes React 19 API changes. ## The Catch Compound components and boolean-prop cleanup, including React 19. Same undeclared-repo LICENSE. Not a performance skill. --- # Vercel React Native Skills > React Native and Expo performance and native-module patterns **Subcategory:** Vercel **License:** MIT **Source:** https://github.com/vercel-labs/agent-skills/blob/main/skills/react-native-skills/SKILL.md --- ## When it is useful React Native and Expo best practices for building performant mobile apps. Use when building React Native components, optimizing list performance, implementing animations, or working with native modules. Triggers on tasks involving React Native, Expo, mobile performance, or native platform APIs. ## The Catch React Native/Expo lists, animations, native modules. Same undeclared-repo LICENSE. Idle on web React. --- # Use Railway > Railway signup, deploy, services, and infrastructure ops **Subcategory:** Other **License:** MIT **Source:** https://github.com/railwayapp/railway-skills/blob/main/plugins/railway/skills/use-railway/SKILL.md --- ## When it is useful Operate Railway infrastructure: sign up for or sign in to a Railway account, create projects, provision services and databases, manage object storage buckets, deploy code, configure infrastructure as code, environments and variables, manage domains, troubleshoot failures, check status and metrics, manage feature flags, set up Railway agent tooling, and query Railway docs. Use this skill whenever the user mentions Railway, feature flags, flag rollout, targeting rules, signing up, creating an account, registering, logging in, deployments, services, environments, buckets, object storage, build failures, agent setup, MCP, or infrastructure operations, even if they don't say "Railway" explicitly. Also invoke this skill when the user asks to be signed up, registered, or onboarded to Railway: do not refuse — drive them through the unauthed `railway up` flow (deploys + signs up on the fly) or `railway login` (which creates new accounts on the fly). ## The Catch False-fires on feature flags, MCP, and signup. Drives `railway up` for signup. A free tier exists. --- # Wrangler > Cloudflare Workers CLI, before you type it **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/wrangler/SKILL.md --- ## When it is useful Cloudflare Workers CLI for deploying, developing, and managing Workers, KV, R2, D1, Vectorize, Hyperdrive, Workers AI, Containers, Queues, Workflows, Pipelines, and Secrets Store. Load before running wrangler commands to ensure correct syntax and best practices. Biases towards retrieval from Cloudflare docs over pre-trained knowledge. ## The Catch Workers CLI for KV/R2/D1/Queues before you type it. Pulls a non-Cloudflare deploy toward wrangler. --- # Workers Best Practices > Production Workers anti-patterns **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/workers-best-practices/SKILL.md --- ## When it is useful Reviews and authors Cloudflare Workers code against production best practices. Load when writing new Workers, reviewing Worker code, configuring wrangler.jsonc, or checking for common Workers anti-patterns (streaming, floating promises, global state, secrets, bindings, observability). Biases towards retrieval from Cloudflare docs over pre-trained knowledge. ## The Catch Production Workers anti-patterns (streaming, globals, bindings). Not a generic JavaScript review. --- # Durable Objects > Stateful coordination on Workers **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/durable-objects/SKILL.md --- ## When it is useful Create and review Cloudflare Durable Objects. Use when building stateful coordination (chat rooms, multiplayer games, booking systems), implementing RPC methods, SQLite storage, alarms, WebSockets, or reviewing DO code for best practices. Covers Workers integration, wrangler config, and testing with Vitest. Biases towards retrieval from Cloudflare docs over pre-trained knowledge. ## The Catch Stateful Durable Objects (SQLite, alarms, WebSockets). False-fires on any “stateful agent” that isn’t a DO. --- # Turnstile Spin > Bot check via Cloudflare Turnstile **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/turnstile-spin/SKILL.md --- ## When it is useful Set up Cloudflare Turnstile end-to-end in a project. Scan the codebase, create the widget via the Cloudflare API, embed it where user requests need bot verification (form submissions, SPA actions, API endpoints, download links, comment or vote submissions, etc.), wire canonical server-side siteverify in the customer's existing backend, validate, and persist the skill. Load this when a user asks to add Turnstile, set up CAPTCHA, protect a form or endpoint from bots, or fix a Turnstile integration. Mirrors developers.cloudflare.com/turnstile/spin. ## The Catch End-to-end Turnstile widget plus siteverify. A generic captcha/bot-check still gets Turnstile. --- # Sandbox Stable > @cloudflare/sandbox on the default tag **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/sandbox-stable/SKILL.md --- ## When it is useful Use when building or changing Cloudflare Sandbox apps on the current stable @cloudflare/sandbox package (default npm tag)—commands, sessions, files, ports, tunnels, terminals, bridge, production, or deprecated-API cleanup while staying on stable. Not for @cloudflare/sandbox@next (use sandbox-next) or for porting to 1.0 (use sandbox-migrate-to-next). ## The Catch @cloudflare/sandbox on the default tag. Wrong sibling if the app is on @next. --- # Sandbox Next > Sandbox SDK 1.0 preview tag **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/sandbox-next/SKILL.md --- ## When it is useful Use when building or changing Cloudflare Sandbox apps on @cloudflare/sandbox@next (Sandbox SDK 1.0 preview)—code execution, AI runners, interpreters, CI-like jobs, terminals, files, mounts, tunnels, preview URLs, lifecycle, or errors. Not for the default stable package (use sandbox-stable) or for porting stable to @next (use sandbox-migrate-to-next). ## The Catch Preview Catch: this is @cloudflare/sandbox@next, not the stable default. Vendor skill for Cloudflare. --- # Cloudflare Email Service > Send and route mail from Workers **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/cloudflare-email-service/SKILL.md --- ## When it is useful Send and receive transactional emails with Cloudflare Email Service (Email Sending + Email Routing). Use when building email sending (Workers binding or REST API), email routing, Agents SDK email handling, or integrating email into any app — Workers, Node.js, Python, Go, etc. Also use for email deliverability, SPF/DKIM/DMARC, wrangler email setup, MCP email tools, or when a coding agent needs to send emails. Even for simple requests like "add email to my Worker" — this skill has critical config details. ## The Catch Workers email send/route and SPF/DKIM. “Add email” on a non-Workers app still steers here. --- # Agents SDK > Stateful agents on Cloudflare Workers **Subcategory:** Cloudflare **License:** Apache-2.0 **Source:** https://github.com/cloudflare/skills/blob/main/skills/agents-sdk/SKILL.md --- ## When it is useful Build AI agents on Cloudflare Workers using the Agents SDK. Load when creating stateful agents, durable workflows, real-time WebSocket apps, scheduled tasks, MCP servers, chat applications, voice agents, or browser automation. Covers Agent class, state management, callable RPC, Workflows, durable execution, queues, retries, observability, and React hooks. Biases towards retrieval from Cloudflare docs over pre-trained knowledge. ## The Catch Cloudflare Agents SDK on Workers (state, RPC, Workflows). “MCP/chat agent” triggers steal from the MCP cluster. --- # Agents.md > This writes AGENTS.md, not a skill **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/agents-md/SKILL.md --- ## When it is useful Creates and maintains concise AGENTS.md and CLAUDE.md project instruction files. Use when asked to create AGENTS.md, update AGENTS.md, maintain agent docs, set up CLAUDE.md, document repository agent conventions, or keep coding-agent instructions minimal and reference-backed. ## The Catch Loud Catch: this IS AGENTS.md / CLAUDE.md, not an agent skill. Do not confuse it with SKILL.md. Vendor skill from Sentry. --- # Claude Settings Audit > permissions.json from the repo shape **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/claude-settings-audit/SKILL.md --- ## When it is useful Analyze a repository to generate recommended Claude Code settings.json permissions. Use when setting up a new project, auditing existing settings, or determining which read-only bash commands to allow. Detects tech stack, build tools, and monorepo structure. ## The Catch Writes Claude Code settings.json allowlists from repo shape. Sentry conventions leak; a generic repo still gets Sentry-shaped permissions. --- # Django Access Review > IDOR and tenant isolation in Django **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/django-access-review/SKILL.md --- ## When it is useful Django access control and IDOR security review. Use when reviewing Django views, DRF viewsets, ORM queries, or any Python/Django code handling user authorization. Trigger keywords: "IDOR", "access control", "authorization", "Django permissions", "object permissions", "tenant isolation", "broken access". ## The Catch IDOR and tenant isolation in Django views/ORM. Idle off Django; not a generic auth review. --- # Django Perf Review > N+1 and queryset audits **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/django-perf-review/SKILL.md --- ## When it is useful Django performance code review. Use when asked to "review Django performance", "find N+1 queries", "optimize Django", "check queryset performance", "database performance", "Django ORM issues", or audit Django code for performance problems. ## The Catch N+1 and queryset audit, not a general perf skill. “Optimize Django” false-fires when the bottleneck isn’t the ORM. --- # Sentry Doc Coauthoring > Structured pass with the human still in it **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/doc-coauthoring/SKILL.md --- ## When it is useful Guide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. ## The Catch Human-in-the-loop doc loop from getsentry/skills. Twin of Anthropic doc-coauthoring; never-fires if the agent just writes the doc. --- # GH Review Requests > Unread review-request notifications **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/gh-review-requests/SKILL.md --- ## When it is useful Fetch unread GitHub notifications for open PRs where review is requested from a specified team or opened by a team member. Use when asked to "find PRs I need to review", "show my review requests", "what needs my review", "fetch GitHub review requests", or "check team review queue". ## The Catch Fetches unread GitHub review-request notifications. Needs gh + a team; an empty inbox looks like “no reviews.” --- # GHA Security Review > Pwn-request and workflow injection **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/gha-security-review/SKILL.md --- ## When it is useful GitHub Actions security review for workflow exploitation vulnerabilities. Use when asked to "review GitHub Actions", "audit workflows", "check CI security", "GHA security", "workflow security review", or review .github/workflows/ for pwn requests, expression injection, credential theft, and supply chain attacks. Exploitation-focused with concrete PoC scenarios. ## The Catch Pwn-request and expression-injection in .github/workflows. Exploitation-shaped; not a green-CI checklist. --- # Iterate PR > CI + review loop until the PR is green **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/iterate-pr/SKILL.md --- ## When it is useful Iterate on a PR until actionable CI passes and high/medium review feedback is addressed. Use for PR CI failures, review feedback, or green-check loops; do not wait for human approval, draft status, or merge gates. ## The Catch Loops CI and review until green, and will not wait for human approval or merge gates. Keeps pushing a draft you wanted parked. --- # PR Writer > Reviewer-facing title and body **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/pr-writer/SKILL.md --- ## When it is useful Create or refresh reviewer-facing PR titles and descriptions. Use when opening a PR, updating its title or body, or preparing branch changes for review. ## The Catch Reviewer-facing title and body only. Will rewrite a PR description you already wrote. --- # Prompt Optimizer > Evals and model-family adapters **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/prompt-optimizer/SKILL.md --- ## When it is useful Creates, optimizes, and iteratively refines agent prompts, system prompts, developer prompts, and reusable prompt templates. Use when asked to improve a prompt, optimize a system prompt, rewrite an agent prompt, tune prompt wording, make a prompt more reliable, port prompts between OpenAI, Claude, or Gemini, or build prompt evals. ## The Catch Prompt evals and OpenAI/Claude/Gemini adapters. Not product copy; will “optimize” a prompt that wasn’t the job. --- # Skill Scanner > Audit a SKILL.md before you install it **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/skill-scanner/SKILL.md --- ## When it is useful Scan agent skills for security issues. Use when asked to "scan a skill", "audit a skill", "review skill security", "check skill for injection", "validate SKILL.md", or assess whether an agent skill is safe to install. Checks for prompt injection, malicious scripts, excessive permissions, secret exposure, and supply chain risks. ## The Catch Audits a SKILL.md for injection and secrets before install. Scanning is not the same as running the skill. --- # Find Bugs > Bugs on this branch, not the whole repo **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/find-bugs/SKILL.md --- ## When it is useful Find bugs, security vulnerabilities, and code quality issues in local branch changes. Use when asked to review changes, find bugs, security review, or audit code on the current branch. ## The Catch Bugs on this branch, not the whole repo. A repo-wide “find bugs” is a false fire. --- # Security Review > OWASP-style pass with confidence scores **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/security-review/SKILL.md --- ## When it is useful Security code review for vulnerabilities. Use when asked to "security review", "find vulnerabilities", "check for security issues", "audit security", "OWASP review", or review code for injection, XSS, authentication, authorization, cryptography issues. Provides systematic review with confidence-based reporting. ## The Catch OWASP-style pass with confidence scores. Overlaps find-bugs and gha-security-review; not Sentry-product config. --- # Document API Endpoint > Sentry @extend_schema and TypedDicts **Subcategory:** Sentry **License:** Apache-2.0 **Source:** https://github.com/getsentry/skills/blob/main/skills/document-api-endpoint/SKILL.md --- ## When it is useful Document and type a Sentry API endpoint. Write or fix @extend_schema decorators, specify response TypedDicts, type request parameters, correct type drift between the declared schema and the runtime response, and validate the generated spec. Use when asked to "document an endpoint", "add OpenAPI docs", "add/fix @extend_schema", "type an endpoint response", "fix the response type", "fix type drift", "reuse a response type", "split an overloaded endpoint", "specify the response schema", "add a TypedDict response", "migrate a legacy api-docs path", "fix a parameter type", or "make an endpoint public" / "promote an endpoint" (promotion is one section here). ## The Catch Sentry @extend_schema and response TypedDicts. Dead outside getsentry API views. --- # Claude Security > In-session scan, then patches you apply **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/claude-security/skills/claude-security/SKILL.md --- ## When it is useful The Claude Security menu — pick a job: scan the codebase (the whole repository or a scoped part of it), scan changes (this branch's or a pull request's diff, or one commit), or suggest patches (findings turned into targeted patch files, each verified by a panel of agents, that you apply when you choose). ## The Catch In-session scan, then patch files you apply. Claude Code plugin layout; not Sentry security-review. --- # Claude Automation Recommender > Hooks, skills, MCP — recommendations only **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/claude-code-setup/skills/claude-automation-recommender/SKILL.md --- ## When it is useful Analyze a codebase and recommend Claude Code automations (hooks, subagents, skills, plugins, MCP servers). Use when user asks for automation recommendations, wants to optimize their Claude Code setup, mentions improving Claude Code workflows, asks how to first set up Claude Code for a project, or wants to know what Claude Code features they should use. ## The Catch Recommends hooks, skills, and MCP for Claude Code — it does not install them. Vendor lock. --- # Agent Development > Claude Code subagent frontmatter **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/plugin-dev/skills/agent-development/SKILL.md --- ## When it is useful This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins. ## The Catch Claude Code subagent frontmatter (when-to-use, tools, colors). “Create an agent” outside plugins is a false fire. --- # Command Development > Slash commands for Claude Code plugins **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/plugin-dev/skills/command-development/SKILL.md --- ## When it is useful This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code. ## The Catch Slash commands and AskUserQuestion inside Claude Code plugins. Not a Unix CLI skill. --- # Hook Development > PreToolUse and friends, in plugins **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/plugin-dev/skills/hook-development/SKILL.md --- ## When it is useful This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks with focus on advanced prompt-based hooks API. ## The Catch PreToolUse/PostToolUse/Stop hooks with ${CLAUDE_PLUGIN_ROOT}. Never-fires without that plugin layout. --- # Plugin Structure > plugin.json at the root, components beside it **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/plugin-dev/skills/plugin-structure/SKILL.md --- ## When it is useful This skill should be used when the user asks to "create a plugin", "scaffold a plugin", "understand plugin structure", "organize plugin components", "set up plugin.json", "use ${CLAUDE_PLUGIN_ROOT}", "add commands/agents/skills/hooks", "configure auto-discovery", or needs guidance on plugin directory layout, manifest configuration, component organization, file naming conventions, or Claude Code plugin architecture best practices. ## The Catch plugin.json scaffold and auto-discovery. First-step only; overlaps the other plugin-dev siblings. --- # Plugin Settings > .local.md config for a plugin **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/plugin-dev/skills/plugin-settings/SKILL.md --- ## When it is useful This skill should be used when the user asks about "plugin settings", "store plugin configuration", "user-configurable plugin", ".local.md files", "plugin state files", "read YAML frontmatter", "per-project plugin settings", or wants to make plugin behavior configurable. Documents the .claude/plugin-name.local.md pattern for storing plugin-specific configuration with YAML frontmatter and markdown content. ## The Catch Per-project .claude/.local.md YAML. “Plugin settings” in an editor is a false fire. --- # Build MCP App > In-chat widgets on an MCP server **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/mcp-server-dev/skills/build-mcp-app/SKILL.md --- ## When it is useful This skill should be used when the user wants to build an "MCP app", add "interactive UI" or "widgets" to an MCP server, "render components in chat", build "MCP UI resources", make a tool that shows a "form", "picker", "dashboard" or "confirmation dialog" inline in the conversation, or mentions "apps SDK" in the context of MCP. Use AFTER the build-mcp-server skill has settled the deployment model, or when the user already knows they want UI widgets. ## The Catch In-chat MCP App widgets after the server exists. Not how to build the server (that’s HOLD build-mcp-server). --- # Build MCPB > Bundle a local MCP with its runtime **Subcategory:** Anthropic **License:** Apache-2.0 **Source:** https://github.com/anthropics/claude-plugins-official/blob/main/plugins/mcp-server-dev/skills/build-mcpb/SKILL.md --- ## When it is useful This skill should be used when the user wants to "package an MCP server", "bundle an MCP", "make an MCPB", "ship a local MCP server", "distribute a local MCP", discusses ".mcpb files", mentions bundling a Node or Python runtime with their MCP server, or needs an MCP server that interacts with the local filesystem, desktop apps, or OS and must be installable without the user having Node/Python set up. ## The Catch Bundles a local MCP plus runtime as .mcpb. Desktop/filesystem MCP; not a hosted-server tutorial. --- # Expo Animation > Reanimated, Gesture Handler, and haptics on device **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-animation/SKILL.md --- ## When it is useful Build animations in React Native and Expo, making the decisions in the order that determines whether they feel right — should it animate, which thread it runs on, which properties, spring or timing, how the gesture hands off, how it degrades. Writes the implementation with Reanimated, Gesture Handler, Expo Router and expo-haptics. Use when animating anything in an Expo app, adding gestures, sheets, screen transitions, press feedback or haptics, or fixing motion that stutters on device. For web animation use `animate`. ## The Catch Reanimated/GH/haptics; wrong for web CSS motion; fights expo-ui (don’t use Reanimated for sheets). --- # Expo App Clip > iOS App Clip target beside the parent Expo app **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-app-clip/SKILL.md --- ## When it is useful Add an iOS App Clip target to an Expo app. Use when the user mentions App Clip, AASA, apple-app-site-association, appclips, smart app banner, or wants to ship a lightweight iOS Clip invoked from a URL alongside their parent app. ## The Catch iOS Clip/AASA only; dead on Android/web. --- # Expo Brownfield > Embed Expo in an existing Swift or Kotlin app **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-brownfield/SKILL.md --- ## When it is useful Integrate Expo and React Native into an existing native iOS or Android app. Use when the user mentions brownfield, embedding React Native in a native app, AAR/XCFramework, or adding Expo to an existing Kotlin/Swift project. Covers both the isolated approach and the integrated approach. ## The Catch embed-in-existing-native only; dead on greenfield Expo. --- # Expo Data Fetching > Fetch, React Query, SWR, and Expo Router loaders **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-data-fetching/SKILL.md --- ## When it is useful Use when implementing or debugging ANY network request, API call, or data fetching. Covers fetch API, React Query, SWR, error handling, caching, offline support, and Expo Router data loaders (`useLoaderData`). ## The Catch “ANY network request” steals generic fetch/SWR outside Expo. --- # Expo Design System > Tokens, variants, and an in-app component library for Expo **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-design-system/SKILL.md --- ## When it is useful Build and maintain a design system inside an Expo app - a reusable theme of design tokens (color, spacing, typography, radius, shadow, motion), reusable component structure with variant/size/state prop conventions, and rules for when to extract a repeated view into a shared component. Use when creating or organizing theme files and design tokens (theme.ts / theme/), extending an existing theme or styling library (NativeWind, Tamagui, Restyle, Unistyles) in its own idiom, standardizing styles so screens (including AI-generated ones) look consistent and polished, building an in-app component library, or auditing an app for design-system drift (hardcoded colors, spacing, fonts). For platform styling specifics (semantic colors, HIG rules, native controls) use expo-native-ui; for folder layout of a new app use expo-project-structure. ## The Catch Expo token/theme kit; twin of frontend-design if you drop the Expo scope. --- # Expo Dev Client > Local and TestFlight Expo development clients **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-dev-client/SKILL.md --- ## When it is useful Build and distribute Expo development clients locally or via TestFlight for internal testing. For production TestFlight releases and store submission, use the eas-app-stores skill. ## The Catch internal dev-client builds; agents mix it with paid eas-app-stores TestFlight. --- # Expo DOM > Web code in a native webview, incrementally **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-dom/SKILL.md --- ## When it is useful Use Expo DOM components to run web code in a webview on native and as-is on web. Migrate web code to native incrementally. For the end-to-end migration of a whole web app, use the expo-web-to-native skill. ## The Catch webview DOM components only; not a full web→native port. --- # Expo Examples > Adapt with-* starters from expo/examples **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-examples/SKILL.md --- ## When it is useful Adapt a canonical `with-*` from expo/examples (~70 integrations) or `npx create-expo --example`. Use when starting from an official Expo example rather than a blank app. Never scaffold on top of an existing app. Distinct from expo-project-structure (new-app folder layout, not example starters). ## The Catch Not full apps — managed single-screen patterns. catalog.md drifts; confirm live list. Never scaffold on top of an existing app. Distinct from expo-project-structure. --- # Expo Migrate Module > Swift Expo module 1.0 DSL → 2.0 macros **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo-experiments/skills/expo-migrate-module/SKILL.md --- ## When it is useful Swift Expo module 1.0 DSL → 2.0 macros (`@ExpoModule`, `@JS`, …). Preserve the JS contract. Use when migrating an existing Swift module off the definition DSL onto Expo Modules API 2.0 macros. Requires expo ≥ 57.0.7. Not for new modules (expo-module), not SDK upgrades (expo-upgrade), not Android/Kotlin. ## The Catch Experiments path, not the main expo plugin. Requires expo ≥ 57.0.7. Not for new modules (expo-module), not SDK upgrades (expo-upgrade), not Android/Kotlin. --- # Expo Module > New Expo Modules API in Swift, Kotlin, and TypeScript **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-module/SKILL.md --- ## When it is useful Guide for creating and writing Expo native modules and views using the Expo Modules API (Swift, Kotlin, TypeScript). Covers module definition DSL, native views, shared objects, config plugins, lifecycle hooks, autolinking, and type system. Use when building or modifying native modules for Expo. Not for migrating an existing Swift module from the definition DSL to the Expo Modules API 2.0 macros; use expo-migrate-module (from the expo-experiments plugin) for that. ## The Catch native module authors (Swift/Kotlin); never-fires on JS-only Expo apps. --- # Expo Native UI > HIG, semantic color, and SF Symbols on Expo screens **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-native-ui/SKILL.md --- ## When it is useful Build beautiful, native-feeling Expo screens. Covers Apple HIG styling, semantic colors, native controls, SF Symbols, media, animations, visual effects, gradients, storage, and responsive layout. For routing and navigation, use the expo-router skill. ## The Catch HIG/SF Symbols screens; look vs @expo/ui primitives (expo-ui). --- # Expo Project Structure > Folder layout for a new Expo Router app **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-project-structure/SKILL.md --- ## When it is useful Folder structure for a new Expo app. Use when scaffolding or laying out a new Expo project with Expo Router, or deciding where a file should live in one. For new projects only — never restructure an existing app to match. ## The Catch new-app folders only; forbids restructuring existing apps. --- # Expo Router > File-based routes, Stack, NativeTabs, and modals **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-router/SKILL.md --- ## When it is useful Navigation and routing for Expo Router. Covers file-based routes, groups and dynamic routes, folder organization, Link with previews and context menus, native Stack, page titles, modals and form sheets, NativeTabs, headers and toolbars, and header search bars. ## The Catch Expo Router file routes only; steals Next.js App Router and React Navigation. --- # Expo UI > @expo/ui SwiftUI and Compose primitives **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-ui/SKILL.md --- ## When it is useful Build native UI with the @expo/ui package: real SwiftUI on iOS and Jetpack Compose on Android. Default to @expo/ui for sheets (BottomSheet), pickers, sliders, toggles, menus, and grouped-form sections — do NOT reach for Reanimated, @gorhom/bottom-sheet, or RN built-in Picker/Switch; use @expo/ui instead. Fall back to RN built-ins only when @expo/ui is missing the component. NOTE: @expo/ui List renders native grouped rows like an iOS Settings screen — it is NOT a virtualized list; use FlatList/FlashList for large datasets. Covers universal components (Host, Column, Row, Button, Text, List, BottomSheet, FieldGroup, Switch, Slider, Picker, Menu), drop-in replacements for RN community libraries, and platform-specific SwiftUI/Jetpack Compose trees. Not for Expo Router navigation, Reanimated, or data fetching. ## The Catch @expo/ui SwiftUI/Compose controls; not FlashList; not expo-native-ui styling. --- # Expo Upgrade > Expo SDK upgrades and dependency fixes **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-upgrade/SKILL.md --- ## When it is useful Guidelines for upgrading Expo SDK versions and fixing dependency issues. ## The Catch Expo SDK bump cookbook; stale after the next SDK; not a generic RN upgrade. --- # Expo Web to Native > Port Next/Vite/CRA to Expo native, screen by screen **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/expo-web-to-native/SKILL.md --- ## When it is useful Migrate an existing web React app to a native iOS/Android app with Expo. Use when the user wants to turn a website into a mobile app, port a Next.js/Vite/CRA React codebase to React Native, reuse web code on native incrementally, or asks how web idioms (the DOM, CSS, React Router, localStorage, window) map to native. This is the end-to-end migration guide; use the `expo-dom` skill for the DOM-component mechanism itself. ## The Catch whole-app Next/Vite/CRA → Expo; not expo-dom. --- # EAS App Stores > EAS build and submit to App Store and Play **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-app-stores/SKILL.md --- ## When it is useful EAS service (paid). Deploy Expo apps to the app stores with EAS - build and submit to the iOS App Store, Google Play Store, and TestFlight, configure eas.json build and submit profiles, manage app versions and build numbers, and publish App Store metadata and ASO. Use whenever the user wants to deploy, release, or ship an app to production or the app stores, is preparing a production build, running eas build or eas submit, shipping to TestFlight, bumping version or build numbers, or setting up store listing metadata. For deploying an Expo website or API routes, use the eas-hosting skill. ## The Catch eas build/submit to stores. Free-tier build minutes; Apple/Google developer accounts still cost. Not local expo-dev-client. --- # EAS Hosting > Expo web and API routes via eas deploy **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-hosting/SKILL.md --- ## When it is useful EAS service (paid). Deploy Expo websites and Expo Router API routes to EAS Hosting - export the web bundle, run eas deploy for production and PR preview URLs, manage environment secrets and custom domains, and work within the Cloudflare Workers runtime. Also covers authoring API routes (+api.ts handlers, HTTP methods, request handling, CORS). Use when deploying an Expo web app or API routes, setting up EAS Hosting, or configuring hosting environments and domains. Not for native builds or store releases - use the eas-app-stores skill for those. ## The Catch Expo web/API routes via eas deploy (CF Workers). Authoring routes is free; production deploys use plan allowance. --- # EAS Observe > expo-observe metrics, 10k MAU free cap **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-observe/SKILL.md --- ## When it is useful EAS service (paid). Use for anything related to EAS Observe - adding `expo-observe` to an Expo project (AppMetricsRoot/ObserveRoot HOC, markInteractive and ObserveInteractiveMarker, the useObserve hook, the Expo Router / React Navigation integrations for per-route metrics, user-defined events via `Observe.logEvent`, error reporting via ObserveErrorBoundary and `Observe.reportError`, and runtime config such as sampleRate and dispatchInDebug), querying via the EAS CLI (`eas observe:metrics-summary`, `observe:metrics`, `observe:routes`, `observe:events`, `observe:session`, `observe:versions`), interpreting the resulting metrics (cold/warm launch, TTR, TTI, navigation cold/warm TTR, update download, and the TTI frameRate/device/network params for triaging slow startups), or shipping an Observe integration inside a third-party package. ## The Catch expo-observe metrics. Free plan up to 10k MAU, then paid. Not generic analytics. --- # EAS Simulator > Remote iOS/Android sim on EAS cloud **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-simulator/SKILL.md --- ## When it is useful EAS service (paid). Run and control a user's app on a remote iOS/Android simulator hosted on EAS cloud. Read before running any `eas simulator:*` commands - it has the current syntax for this experimental API. Use whenever the user needs a simulator they can't run locally - 'run my app on a cloud simulator', 'use eas simulator to run/install/screenshot my app', 'I'm on Linux/Cursor and need an iOS device', 'no sim on this box / headless CI', 'let an agent click through my app and screenshot it', 'test my dev build on a remote sim with live reload', 'stream a sim to my browser' - even when they don't say 'EAS Simulator' or 'cloud'. On a host WITHOUT a local simulator (Linux, CI, cloud sandbox) it's the default; on macOS, do NOT auto-trigger for a plain 'run on the simulator' - use it only for a cloud/remote/shareable sim, an iOS version they lack, or an agent-driven session. NOT for local sims (expo run:ios, Xcode, Android Studio), EAS Build/Update, web preview, or physical devices. ## The Catch Remote iOS/Android sim on EAS cloud, experimental API. Not local sim. --- # EAS Update > OTA JS/asset updates via expo-updates **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-update/SKILL.md --- ## When it is useful OTA JS/asset updates via expo-updates + EAS CLI (`eas update`, channels/branches/runtime). Use when publishing over-the-air JavaScript and asset updates to an Expo app without a new native store binary. Distinct from eas-update-insights (health/adoption of published rollouts, not publish) and eas-app-stores (native store binaries). ## The Catch EAS service, costs apply (Free plan has update/bandwidth/storage allowances). Distinct from live eas-update-insights (health/adoption, not publish) and eas-app-stores (native store binaries). --- # EAS Update Insights > Health of published EAS Update rollouts **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-update-insights/SKILL.md --- ## When it is useful EAS service (paid). Check the health of published EAS Update: crash rates, install/launch counts, unique users, payload size, and the split between embedded and OTA users per channel. Use when the user asks how an update is performing, whether a rollout is healthy, how many users are on the embedded build vs OTA, or wants to gate CI on update health. ## The Catch Health of published EAS Update rollouts. Idle if you don’t OTA. --- # EAS Workflows > .eas/workflows YAML that burns compute minutes **Subcategory:** Expo **License:** MIT **Source:** https://github.com/expo/skills/blob/main/plugins/expo/skills/eas-workflows/SKILL.md --- ## When it is useful EAS service (paid). Helps understand and write EAS workflow YAML files for Expo projects. Use this skill when the user asks about CI/CD or workflows in an Expo or EAS context, mentions .eas/workflows/, or wants help with EAS build pipelines or deployment automation. ## The Catch .eas/workflows YAML. Each job burns build/compute minutes. False-fire on generic CI. --- # Supabase Postgres Best Practices > Load before any Postgres schema or SQL **Subcategory:** Supabase **License:** MIT **Source:** https://github.com/supabase/agent-skills/blob/main/skills/supabase-postgres-best-practices/SKILL.md --- ## When it is useful Postgres best practices maintained by Supabase, for Postgres running anywhere. Load this skill BEFORE writing or changing anything that lives in a Postgres database: creating or altering tables and columns (including choosing column types), schema design, migrations and declarative schema files, RLS policies and the tests that verify them, indexes, triggers, database functions, queues and scheduled jobs (pg_cron, pgmq), vector/semantic search (pgvector), and restoring dumps (pg_restore) or importing data. Also load it when diagnosing slow queries, high CPU, timeouts, EXPLAIN plans, connection exhaustion, locking, bloat, or rows visible to the wrong user or tenant. This is not just a performance guide — schema, migration, security, and SQL authoring tasks need these rules too, even for a one-column change or a single query. ## The Catch Load BEFORE any Postgres schema/SQL. Always-on for SQL, even off Supabase. --- # Supabase > Any Supabase product; changelog-first vendor skill **Subcategory:** Supabase **License:** MIT **Source:** https://github.com/supabase/agent-skills/blob/main/skills/supabase/SKILL.md --- ## When it is useful Use when doing ANY task involving Supabase. Triggers: Supabase products (Database, Auth, Edge Functions, Realtime, Storage, Vectors, Cron, Queues); client libraries and SSR integrations (supabase-js, @supabase/ssr) in Next.js, React, SvelteKit, Astro, Remix; auth issues (login, logout, sessions, JWT, cookies, getSession, getUser, getClaims, RLS); Supabase CLI or MCP server; schema changes, migrations, declarative schemas, security audits, Postgres extensions (pg_graphql, pg_cron, pg_vector); debugging and troubleshooting errors or unexpected behavior on Supabase projects (HTTP errors, Postgres errors, RLS surprises, permission denied, schema cache issues, timeouts, Edge Function crashes, Realtime drops, Storage failures) and reading or querying logs (Logs Explorer, ClickHouse). ## The Catch ANY Supabase product. Steals the Postgres skill. Changelog-first; vendor lock. --- # Prisma CLI > ORM CLI: init, generate, migrate, studio **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-cli/SKILL.md --- ## When it is useful Prisma ORM CLI commands reference covering init, generate, migrate, db, dev, complete, studio, validate, format, debug, and mcp. Use for ORM/database CLI workflows, not the Prisma Platform CLI. Triggers on "prisma init", "prisma generate", "prisma migrate", "prisma db", "prisma complete", "prisma studio", "prisma mcp". ## The Catch ORM CLI (init/generate/migrate/studio), not Platform CLI. --- # Prisma Client API > findMany, create, $transaction **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-client-api/SKILL.md --- ## When it is useful Prisma Client API reference covering model queries, filters, operators, and client methods. Use when writing database queries, using CRUD operations, filtering data, or configuring Prisma Client. Triggers on "prisma query", "findMany", "create", "update", "delete", "$transaction". ## The Catch findMany/create/$transaction. Idle if you’re not writing Prisma Client. --- # Prisma Database Setup > Wire Prisma to Postgres, MySQL, SQLite, Mongo **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-database-setup/SKILL.md --- ## When it is useful Guides for configuring Prisma with different database providers (PostgreSQL, MySQL, SQLite, MongoDB, etc.). Use when setting up a new project, changing databases, or troubleshooting connection issues. Triggers on "configure postgres", "connect to mysql", "setup mongodb", "sqlite setup". ## The Catch Wire Prisma to Postgres/MySQL/SQLite/Mongo. Not Prisma Postgres hosting. --- # Prisma Driver Adapter Implementation > ORM 7 SQL driver adapters **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-driver-adapter-implementation/SKILL.md --- ## When it is useful Required reference for Prisma ORM 7 SQL driver adapter work. Use when implementing or modifying adapters, adding database drivers, or touching SqlDriverAdapter, Transaction, savepoint, result mapping, or DriverAdapterError behavior. Covers current transaction lifecycle, optional savepoint hooks, original database-error preservation, and verification. ## The Catch ORM 7 SQL driver adapters. Never-fires for app CRUD. --- # Prisma MongoDB Upgrade > Mongo on v6 has no v7 path **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-mongodb-upgrade/SKILL.md --- ## When it is useful Decision and migration guide for Prisma ORM MongoDB projects on v6, which have no upgrade path to v7. Use when a MongoDB project asks about upgrading Prisma, when "upgrade to prisma 7" comes up in a project with provider = "mongodb", or when evaluating a move to Prisma Next. Triggers on "upgrade prisma mongodb", "prisma 7 mongodb", "mongodb prisma migration", "prisma next mongodb". ## The Catch Mongo on v6 has no v7 path. Migrate-only. --- # Prisma Upgrade v7 > v6→v7 breaking changes **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-upgrade-v7/SKILL.md --- ## When it is useful Complete migration guide from Prisma ORM v6 to v7 covering all breaking changes. Use when upgrading Prisma versions, encountering v7 errors, or migrating existing projects. Triggers on "upgrade to prisma 7", "prisma 7 migration", "prisma-client generator", "driver adapter required". ## The Catch v6→v7 breaking changes. Version-gate; will go stale. --- # Prisma Postgres > Hosted Prisma Postgres, not OSS Postgres **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-postgres/SKILL.md --- ## When it is useful Prisma Postgres setup and operations guidance across Console, create-db CLI, Management API, and Management API SDK. Use when creating Prisma Postgres databases, working in Prisma Console, provisioning with create-db/create-pg/create-postgres, or integrating programmatic provisioning with service tokens or OAuth. ## The Catch Hosted Prisma Postgres. Vendor; Free $0/mo with hard cap (prisma.io/postgres). Not OSS Postgres. --- # Prisma Compute > Prisma Compute deploy, 1M requests/mo free **Subcategory:** Prisma **License:** MIT **Source:** https://github.com/prisma/skills/blob/main/prisma-compute/SKILL.md --- ## When it is useful Prisma Compute deployment and hosting guide. Use whenever the user mentions Prisma Compute, `prisma.compute.ts`, `defineComputeConfig`, deploying or hosting a Prisma app, `@prisma/cli app deploy`, `compute:deploy`, `create-prisma --deploy`, `PRISMA_SERVICE_TOKEN`, Compute auth/workspaces, apps/deployments/build logs/domains, localhost vs `0.0.0.0`, deploy port binding, or framework deploy readiness for Hono, Elysia, Next.js, TanStack Start, Astro, Nuxt, Svelte, Nest, Turborepo, or custom/prebuilt artifacts. ## The Catch Prisma Compute deploy. Vendor; Free includes 1M requests/mo. Not a generic host. --- # Threat Model Generation > STRIDE threat model for a repo **Subcategory:** Factory **License:** MIT **Source:** https://github.com/factory-ai/skills/blob/main/skills/threat-model-generation/SKILL.md --- ## When it is useful Generate a STRIDE-based security threat model for a repository. Use when setting up security monitoring, after architecture changes, or for security audits. ## The Catch STRIDE doc; not sentry security-review. --- # Firecrawl Agent > Multi-page schema extract via Firecrawl agent **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-agent/SKILL.md --- ## When it is useful Autonomous multi-page extraction into structured JSON. Use when the user wants website data matching a schema — pricing tiers, product listings — beyond a single-page scrape. ## The Catch multi-page schema extract; credits; not a generic crawler. Self-host via FIRECRAWL_API_URL is the free path; cloud needs a key. --- # Firecrawl Crawl > Bulk crawl; no keyless free tier **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-crawl/SKILL.md --- ## When it is useful Bulk-extract many pages from one site or section. Use for "crawl", "everything under /docs", or content spanning linked pages. ## The Catch LOUD: skill text: crawl has no keyless free tier; credits per page; auth required. --- # Firecrawl Build Scrape > One-URL /scrape in app code **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/build/firecrawl-build-scrape/SKILL.md --- ## When it is useful Integrate Firecrawl `/scrape` into product code for single-page extraction. Use when an app already has a URL and needs markdown, HTML, links, screenshots, metadata, or structured page output. Prefer this skill over broader crawl patterns when the feature is page-level. ## The Catch one-URL scrape in app code. Self-host option. Not crawl. --- # Firecrawl Build Search > Firecrawl /search in product code **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/build/firecrawl-build-search/SKILL.md --- ## When it is useful Integrate Firecrawl `/search` into product code and agent workflows. Use when an app needs discovery before extraction, when the feature starts with a query instead of a URL, or when the system should search the web and optionally hydrate result content. ## The Catch Firecrawl /search; not a generic web-search skill. --- # Firecrawl Build Interact > Clicks and forms after scrape **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/build/firecrawl-build-interact/SKILL.md --- ## When it is useful Integrate Firecrawl `/interact` into product code for dynamic pages and browser actions after scraping. Use when a feature needs clicks, form fills, pagination, authentication-aware flows, or other multi-step interactions that plain `/scrape` cannot complete. ## The Catch clicks/forms after scrape. Overlaps shipped webapp-testing and Browserbase browser. --- # Hugging Face Papers > HF/arXiv URL to markdown **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-papers/SKILL.md --- ## When it is useful Look up and read Hugging Face paper pages in markdown, and use the papers API for structured metadata such as authors, linked models/datasets/spaces, Github repo and project page. Use when the user shares a Hugging Face paper page URL, an arXiv URL or ID, or asks to summarize, explain, or analyze an AI research paper. ## The Catch HF/arXiv URL → markdown. Papers only. --- # Browser > NL browse CLI; local Chrome or Browserbase **Subcategory:** Browserbase **License:** MIT (frontmatter; no LICENSE file) **Source:** https://github.com/browserbase/skills/blob/main/skills/browser/SKILL.md --- ## When it is useful Automate web browser interactions using natural language via CLI commands. Use when the user asks to browse websites, navigate web pages, extract data from websites, take screenshots, fill forms, click buttons, or interact with web applications. Supports remote Browserbase sessions with Browserbase Identity, Verified browsers, automatic CAPTCHA solving, and residential proxies — ideal for protected websites and JavaScript-heavy pages. ## The Catch NL browse CLI. Local Chrome is the free path; remote Browserbase needs API key. Loud: no repo LICENSE (frontmatter MIT only). --- # Browser Trace > CDP/screenshot debug of a failed browse **Subcategory:** Browserbase **License:** MIT (frontmatter; no LICENSE file) **Source:** https://github.com/browserbase/skills/blob/main/skills/browser-trace/SKILL.md --- ## When it is useful Capture a full DevTools-protocol trace of any browser automation — CDP firehose, screenshots, and DOM dumps — then bisect the stream into per-page searchable buckets. Use when the user wants to debug a failed run, audit network/console/DOM activity, attach a trace to an in-progress session, or feed structured per-page summaries back into an agent loop so its next iteration learns from the last one. ## The Catch CDP/screenshot debug of a failed browse. Dead without browser. --- # Browser to API > Trace to OpenAPI 3.1 from XHR **Subcategory:** Browserbase **License:** MIT (frontmatter; no LICENSE file) **Source:** https://github.com/browserbase/skills/blob/main/skills/browser-to-api/SKILL.md --- ## When it is useful Turn a website's observable HTTP traffic into a best-effort OpenAPI 3.1 spec by analyzing a `browser-trace` capture. Use when the user wants to discover/extract API endpoints from a browser session, build an OpenAPI doc from network traffic, or document a third-party site's XHR/fetch surface for client integration. ## The Catch trace → OpenAPI 3.1 from XHR. Needs browser-trace first; not an exploit skill. --- # Transformers.js > Transformers.js in browser and Node **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/transformers-js/SKILL.md --- ## When it is useful Use Transformers.js to run state-of-the-art machine learning models directly in JavaScript/TypeScript. Supports NLP (text classification, translation, summarization), computer vision (image classification, object detection), audio (speech recognition, audio classification), and multimodal tasks. Works in browsers and server-side runtimes (Node.js, Bun, Deno) with WebGPU/WASM using pre-trained models from Hugging Face Hub. ## The Catch Transformers.js in browser/Node. Weight downloads and RAM are the limit. --- # Hugging Face Local Models > GGUF + llama.cpp on CPU/Metal/CUDA/ROCm **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-local-models/SKILL.md --- ## When it is useful Use to select models to run locally with llama.cpp and GGUF on CPU, Mac Metal, CUDA, or ROCm. Covers finding GGUFs, quant selection, running servers, exact GGUF file lookup, conversion, and OpenAI-compatible local serving. ## The Catch GGUF + llama.cpp on CPU/Metal/CUDA/ROCm. Not a cloud-Jobs trainer. --- # Hugging Face Datasets > Hub Dataset Viewer API **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-datasets/SKILL.md --- ## When it is useful Use this skill for Hugging Face Dataset Viewer API workflows that fetch subset/split metadata, paginate rows, search text, apply filters, download parquet URLs, and read size or statistics. ## The Catch Hub Dataset Viewer API. Not local datasets training prep. (frontmatter nonstandard; file 200.) --- # Hugging Face Gradio > Gradio apps in Python **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-gradio/SKILL.md --- ## When it is useful Build Gradio web UIs and demos in Python. Use when creating or editing Gradio apps, components, event listeners, layouts, or chatbots. ## The Catch Gradio apps only. Not a general frontend skill. --- # Hugging Face Trackio > Trackio metrics, not W&B **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-trackio/SKILL.md --- ## When it is useful Track and visualize ML training experiments with Trackio. Use when logging metrics during training (Python API), firing alerts for training diagnostics, or retrieving/analyzing logged metrics (CLI). Supports real-time dashboard visualization, alerts with webhooks, HF Space syncing, and JSON output for automation. ## The Catch Trackio metrics. Dead if the project uses W&B/MLflow. --- # Hugging Face Community Evals > Local inspect-ai and lighteval **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-community-evals/SKILL.md --- ## When it is useful Run evaluations for Hugging Face Hub models using inspect-ai and lighteval on local hardware. Use for backend selection, local GPU evals, and choosing between vLLM / Transformers / accelerate. Not for HF Jobs orchestration, model-card PRs, .eval_results publication, or community-evals automation. ## The Catch local inspect-ai/lighteval. Not HF Jobs. --- # Hugging Face Paper Publisher > Create and claim HF paper pages **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-paper-publisher/SKILL.md --- ## When it is useful Publish and manage research papers on Hugging Face Hub. Supports creating paper pages, linking papers to models/datasets, claiming authorship, and generating professional markdown-based research articles. ## The Catch create/claim HF paper pages. Distinct from huggingface-papers (reader). --- # Hugging Face Vision Trainer > Detection/SAM on HF Jobs GPUs **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-vision-trainer/SKILL.md --- ## When it is useful Trains and fine-tunes vision models for object detection (D-FINE, RT-DETR v2, DETR, YOLOS), image classification (timm models — MobileNetV3, MobileViT, ResNet, ViT/DINOv3 — plus any Transformers classifier), and SAM/SAM2 segmentation using Hugging Face Transformers on Hugging Face Jobs cloud GPUs. Covers COCO-format dataset preparation, Albumentations augmentation, mAP/mAR evaluation, accuracy metrics, SAM segmentation with bbox/point prompts, DiceCE loss, hardware selection, cost estimation, Trackio monitoring, and Hub persistence. Use when users mention training object detection, image classification, SAM, SAM2, segmentation, image matting, DETR, D-FINE, RT-DETR, ViT, timm, MobileNet, ResNet, bounding box models, or fine-tuning vision models on Hugging Face Jobs. ## The Catch detection/SAM on HF Jobs GPUs. Loud Catch: cloud-GPU costs apply. --- # Playwright > playwright-cli from the terminal **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/playwright/SKILL.md --- ## When it is useful Use when the task requires automating a real browser from the terminal (navigation, form filling, snapshots, screenshots, data extraction, UI-flow debugging) via `playwright-cli` or the bundled wrapper script. ## The Catch playwright-cli from the terminal. Twin of shipped webapp-testing (CLI wrapper, not the library). --- # Jupyter Notebook > .ipynb only, not a Python app skill **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/jupyter-notebook/SKILL.md --- ## When it is useful Use when the user asks to create, scaffold, or edit Jupyter notebooks (`.ipynb`) for experiments, explorations, or tutorials; prefer the bundled templates and run the helper script `new_notebook.py` to generate a clean starting notebook. ## The Catch .ipynb only. Not a Python app skill. --- # GH Fix CI > Failing Actions on the current PR **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/gh-fix-ci/SKILL.md --- ## When it is useful Use when a user asks to debug or fix failing GitHub PR checks that run in GitHub Actions; use `gh` to inspect checks and logs, summarize failure context, draft a fix plan, and implement only after explicit approval. Treat external providers (for example Buildkite) as out of scope and report only the details URL. ## The Catch failing Actions on the current PR; implement only after explicit approval. Buildkite out of scope. --- # GH Address Comments > Review comments on the current-branch PR **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/gh-address-comments/SKILL.md --- ## When it is useful Help address review/issue comments on the open GitHub PR for the current branch using gh CLI; verify gh auth first and prompt the user to authenticate if not logged in. ## The Catch review comments on the current-branch PR via gh. Needs gh auth. --- # Yeet > Stage, commit, push, and open a PR **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/yeet/SKILL.md --- ## When it is useful Use only when the user explicitly asks to stage, commit, push, and open a GitHub pull request in one flow using the GitHub CLI (`gh`). ## The Catch stage+commit+push+PR in one flow. Loud: only when the user explicitly asks; dangerous if it false-fires. --- # Screenshot > OS-level capture, not web page shots **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/screenshot/SKILL.md --- ## When it is useful Use when the user explicitly asks for a desktop or system screenshot (full screen, specific app or window, or a pixel region), or when tool-specific capture capabilities are unavailable and an OS-level capture is needed. ## The Catch OS-level capture. Not web page shots. --- # OpenAI Docs > Current OpenAI/Codex docs; twin of claude-api **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/openai-docs/SKILL.md --- ## When it is useful Use when the user asks how to build with OpenAI products or APIs, asks about Codex itself or choosing Codex surfaces, needs up-to-date official documentation with citations, help choosing the latest model for a use case, or model upgrade and prompt-upgrade guidance; use OpenAI docs MCP tools for non-Codex docs questions, use the Codex manual helper first for broad Codex self-knowledge, and restrict fallback browsing to official OpenAI domains. ## The Catch current OpenAI/Codex docs. Loud vendor; twin of claude-api. --- # Netlify Deploy > npx netlify; Netlify says build free **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/netlify-deploy/SKILL.md --- ## When it is useful Deploy web projects to Netlify using the Netlify CLI (`npx netlify`). Use when the user asks to deploy, host, publish, or link a site/repo on Netlify, including preview and production deploys. ## The Catch npx netlify. Vendor; netlify.com/pricing: “Build free.” --- # Render Deploy > render.yaml; Render says deploy for free **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/render-deploy/SKILL.md --- ## When it is useful Deploy applications to Render by analyzing codebases, generating render.yaml Blueprints, and providing Dashboard deeplinks. Use when the user wants to deploy, host, publish, or set up their application on Render's cloud platform. ## The Catch render.yaml. Vendor; render.com/pricing: “Deploy for free.” --- # Linear > Issues via Linear; Free $0, needs auth **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/linear/SKILL.md --- ## When it is useful Manage issues, projects & team workflows in Linear. Use when the user wants to read, create or updates tickets in Linear. ## The Catch issues via Linear. Vendor; linear.app/pricing Free $0. Dead without Linear auth. --- # Notion Knowledge Capture > Chats into a Notion wiki or how-to **Subcategory:** Notion **License:** MIT **Source:** https://github.com/openai/plugins/blob/main/plugins/notion/skills/notion-knowledge-capture/SKILL.md --- ## When it is useful Capture conversations and decisions into structured Notion pages; use when turning chats/notes into wiki entries, how-tos, decisions, or FAQs with proper linking. ## The Catch chats → Notion wiki/how-to. Vendor; notion.so/pricing has a Free plan. Dead without Notion MCP. --- # Upstash Redis Start > Scratch Redis, 72h, no signup **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-redis-start/SKILL.md --- ## When it is useful Provision a zero-config, no-signup, temporary Upstash Redis database for an AI agent with a single POST to https://upstash.com/start-redis, with no account, API key, or SDK setup required. Use when an agent needs scratch Redis right now and the user has not provided credentials, for short-term memory across tool calls, conversation history, a sub-agent work queue, ranked recall, or a quick prototype or demo. Covers idempotent creation and re-fetching credentials, calling the database through the body-style REST API or the official SDKs, and telling the user how to claim it. The database lives 3 days unless the user claims it; not for production data, PII, or secrets. ## The Catch scratch Redis, no signup. Temporary; not production. Free path: start-redis 72h no-signup. --- # Upstash Redis JS > @upstash/redis HTTP client, not ioredis **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-redis-js/SKILL.md --- ## When it is useful Work with the @upstash/redis TypeScript/JavaScript SDK, a serverless HTTP-based Redis client for Next.js, Vercel, Cloudflare Workers, edge runtimes, and Node.js. Use when adding a cache (cache-aside, write-through, TTL and expiration strategies), session storage and user sessions, a key-value store, leaderboards and rankings with sorted sets, counters, distributed locks, queues with lists, streams and consumer groups, JSON documents, pipelines and MULTI/EXEC transactions, Lua scripting, read replicas, or full-text search, typo-tolerant search, facets, and aggregations with Upstash Redis Search (different from regular FT.SEARCH; also available for TCP clients via @upstash/search-redis and @upstash/search-ioredis). Also use when migrating from ioredis or node-redis, when a Redis connection is needed from a serverless function without connection pooling, when integrating @upstash/ratelimit, or when the user says Redis cache, KV store, session store, serverless Redis, or Upstash Redis. Supports automatic serialization/deserialization of JavaScript types. ## The Catch @upstash/redis HTTP client. Not ioredis/node-redis. Free $0 on upstash.com/pricing. --- # Upstash Ratelimit JS > Rate limit on top of Upstash Redis **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-ratelimit-js/SKILL.md --- ## When it is useful Rate limiting for serverless and edge apps with the @upstash/ratelimit TypeScript/JavaScript SDK backed by Upstash Redis. Use when adding a rate limiter or throttling to an API route, Next.js middleware, Vercel Edge, Cloudflare Workers, or any HTTP endpoint; returning 429 Too Many Requests; choosing between fixed window, sliding window, and token bucket algorithms; limiting per user, IP, API key, or tenant with prefixes and custom keys; protecting login, signup, form, or AI endpoints from abuse, bots, and brute force; using deny lists, ephemeral caching, analytics, timeouts, and multi-region rate limits; or estimating the Redis command cost of rate limiting. Also use when the user says rate limit, rate-limiting, throttle, quota, request limits, or traffic protection. ## The Catch needs Upstash Redis underneath. Free $0 on upstash.com/pricing. --- # Upstash QStash JS > HTTP queue and schedules on QStash **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-qstash-js/SKILL.md --- ## When it is useful Work with the @upstash/qstash TypeScript/JavaScript SDK, an HTTP-based message queue, task scheduler, and background job system for serverless and edge runtimes (Next.js, Vercel, Cloudflare Workers, Deno, Node.js). Use when publishing messages to HTTP endpoints or URL groups, running background jobs without a long-running worker process, scheduling with cron expressions, delaying messages, building FIFO queues with parallelism and flow control, configuring retries and callbacks, handling a dead letter queue (DLQ), deduplicating messages, fanning out to multiple endpoints, verifying QStash webhook signatures (Next.js App Router, Pages Router, and Edge Runtime), running a local QStash dev server, or migrating regions. Also use when the user asks for a serverless cron job, async task queue, job scheduler, delayed delivery, webhook delivery with retries, or event-driven messaging between services. ## The Catch HTTP queue/schedules. Vendor; Free $0 on pricing. --- # Upstash Workflow JS > Durable workflows on QStash **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-workflow-js/SKILL.md --- ## When it is useful Work with the @upstash/workflow TypeScript/JavaScript SDK for durable, long-running workflows in serverless functions, multi-step processes that survive timeouts, retries, and restarts (built on QStash). Use when defining a workflow endpoint with serve(), running steps with context.run, sleeping for minutes to days without holding a function open, calling external APIs with context.call, waiting for an external event or webhook, invoking other workflows, configuring retries, failure callbacks, and a DLQ, controlling concurrency, rate, and parallelism, triggering, cancelling, or inspecting runs with the Workflow client, building AI agents and orchestrators, human-in-the-loop approvals, realtime updates, local development with the QStash dev server, adding middleware, or migrating workflows safely. Also use when the user asks for durable execution, step functions, saga or orchestration patterns, background jobs with checkpoints, or long-running tasks on Vercel, Next.js, Cloudflare Workers, or other serverless platforms. ## The Catch durable workflows on QStash. Distinct from qstash-js, not a twin dump. --- # Upstash Vector JS > Upstash Vector, not pgvector **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-vector-js/SKILL.md --- ## When it is useful Work with the @upstash/vector TypeScript/JavaScript SDK, a serverless vector database for embeddings, similarity search, semantic search, and RAG (retrieval-augmented generation). Use when upserting, querying, fetching, ranging, or deleting vectors, upserting raw text against an index with a built-in embedding model, choosing dense, sparse, or hybrid indexes, filtering by metadata, organizing data with namespaces, running resumable queries, or connecting Upstash Vector to an AI or LLM application. Also use when the user asks for a vector store, vector search, nearest-neighbor or kNN search, embeddings storage, semantic cache, recommendations or similarity features, or a hosted vector index that needs no infrastructure. ## The Catch not pgvector. Free $0 on upstash.com/pricing. --- # Upstash Search JS > Upstash Search vendor SDK **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-search-js/SKILL.md --- ## When it is useful Work with the @upstash/search TypeScript/JavaScript SDK, a serverless full-text and semantic search database with built-in reranking. Use when adding search to an app or site, creating a search index, upserting documents with searchable content and filterable metadata, running keyword, semantic, or hybrid search queries, reranking results, filtering with SQL-like or structured filter syntax, paginating with range, fetching or deleting documents, resetting an index, or checking index info. Also use when the user asks for site search, product, document, or knowledge-base search, or a managed search service that needs no cluster to run. ## The Catch Upstash Search vendor. Free $0 on upstash.com/pricing. --- # Upstash CLI > upstash CLI; needs API token **Subcategory:** Upstash **License:** MIT **Source:** https://github.com/upstash/skills/blob/main/skills/upstash-cli/SKILL.md --- ## When it is useful Run the Upstash CLI (`upstash`) against the Upstash Developer API for Redis, Vector, Search, QStash, and teams, with non-interactive commands and JSON output for scripts, CI, and agents. Use when creating, listing, renaming, or deleting Redis databases, changing plans, regions, TLS, eviction, auto-upgrade, or budgets, managing backups, running Redis commands with `upstash redis exec`, creating or inspecting Vector and Search indexes, managing QStash instances and tokens, managing team members, reading usage stats, or automating any Upstash account operation from the terminal. Also use when the user asks how to provision or manage Upstash resources without the console. ## The Catch needs API token; CI/agent oriented. --- # PlanetScale pscale CLI Automation > Headless pscale with --format json **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-pscale-cli-automation/SKILL.md --- ## When it is useful Use the PlanetScale CLI (pscale) from automated agents with --format json, auth check, pscale sql, and per-command --force. Run before other PlanetScale skills when driving pscale directly. Use when the user asks to automate pscale, run CLI commands headless, or verify pscale auth from an agent. ## The Catch LOUD: planetscale.com/pricing has no verified $0 hobby (calculator from ~$15; “first 10 GB free” is storage add-on language, not a catalog free tier). --- # PlanetScale Postgres Safety Review > PS Postgres only; not the Supabase skill **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-postgres-safety-review/SKILL.md --- ## When it is useful Review PlanetScale Postgres for Traffic Control, query tags, roles, pg_strict, backups/PITR, private connectivity, webhooks, branches, and safe agent operation. ## The Catch PS Postgres only; not the Supabase Postgres skill. LOUD: no verified $0 PlanetScale hobby. --- # PlanetScale Vitess Safety Review > Vitess branch safety review **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-vitess-safety-review/SKILL.md --- ## When it is useful Review a PlanetScale Vitess database for safe migrations, deploy requests, schema recommendations, Insights, webhooks, and operational safety. ## The Catch Vitess branch only. LOUD: no verified $0 PlanetScale hobby. --- # PlanetScale Query Insights and Tags > PlanetScale query insights and tags **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-query-insights-and-tags/SKILL.md --- ## When it is useful Use PlanetScale Insights and SQLCommenter-style query tags to attribute database load, identify risky queries, and prepare safe Traffic Control or schema recommendations. ## The Catch PlanetScale query insights/tags. LOUD: no verified $0 PlanetScale hobby. --- # PlanetScale Schema Recommendations Agent Loop > Triage PS schema recs; does not apply prod **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-schema-recommendations-agent-loop/SKILL.md --- ## When it is useful Safely triage PlanetScale schema recommendations and turn them into reviewed branches, migrations, issues, or pull requests without applying production changes. ## The Catch never-fires without PS schema recs; does not apply prod. LOUD: no verified $0 PlanetScale hobby. --- # PlanetScale MCP Agent Operating Model > PS MCP policy; dead without PS MCP **Subcategory:** PlanetScale **License:** MIT **Source:** https://github.com/planetscale/skills/blob/main/planetscale-mcp-agent-operating-model/SKILL.md --- ## When it is useful Configure safe agent behavior around PlanetScale MCP, Insights, schema recommendations, and repository work without autonomous production mutation. ## The Catch policy; dead without PS MCP. LOUD: no verified $0 PlanetScale hobby. --- # WorkOS > WorkOS auth; Free $0/mo, dead without WorkOS **Subcategory:** WorkOS **License:** MIT **Source:** https://github.com/workos/skills/blob/main/plugins/workos/skills/workos/SKILL.md --- ## When it is useful Use when the user asks for a WorkOS docs URL, term, or dashboard field (Sign-in endpoint, initiate_login_uri, Redirect URI, `WORKOS_*` env vars), or is implementing, debugging, or migrating WorkOS — AuthKit, SSO/SAML, Directory Sync, RBAC, FGA, MFA, Vault, Audit Logs, Admin Portal, Pipes (Connected Apps), Feature Flags, Radar (bot/fraud detection), webhooks, Custom Domains, running the `workos` CLI in agent or sandbox sessions (`WORKOS_MODE`, `workos doctor`), or migrating from Auth0, Clerk, Cognito, Firebase, Supabase, Stytch, Descope, or Better Auth. Also triggers on @workos-inc/* imports. ## The Catch WorkOS auth. Vendor; workos.com/pricing Free $0/mo. Dead without WorkOS. --- # ASP.NET Core > .NET-only ASP.NET Core, Blazor, Minimal APIs **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/aspnet-core/SKILL.md --- ## When it is useful Build, review, refactor, or architect ASP.NET Core web applications using current official guidance for .NET web development. Use when working on Blazor Web Apps, Razor Pages, MVC, Minimal APIs, controller-based Web APIs, SignalR, gRPC, middleware, dependency injection, configuration, authentication, authorization, testing, performance, deployment, or ASP.NET Core upgrades. ## The Catch .NET-only. Never-fires outside ASP.NET Core / Blazor / Minimal APIs. --- # ChatGPT Apps > ChatGPT Apps SDK with MCP widget **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/chatgpt-apps/SKILL.md --- ## When it is useful Build, scaffold, refactor, and troubleshoot ChatGPT Apps SDK applications that combine an MCP server and widget UI. Use when Codex needs to design tools, register UI resources, wire the MCP Apps bridge or ChatGPT compatibility APIs, apply Apps SDK metadata or CSP or domain settings, or produce a docs-aligned project scaffold. Prefer a docs-first workflow by invoking the openai-docs skill or OpenAI developer docs MCP tools before generating code. ## The Catch ChatGPT Apps SDK (MCP + widget). Not generic MCP — pulls you into ChatGPT Apps. --- # CLI Creator > Codex companion CLI from OpenAPI or curl **Subcategory:** OpenAI **License:** Apache-2.0 **Source:** https://github.com/openai/skills/blob/main/skills/.curated/cli-creator/SKILL.md --- ## When it is useful Build a composable CLI for Codex from API docs, an OpenAPI spec, existing curl examples, an SDK, a web app, an admin tool, or a local script. Use when the user wants Codex to create a command-line tool that can run from any repo, expose composable read/write commands, return stable JSON, manage auth, and pair with a companion skill. ## The Catch Codex companion CLI from OpenAPI/curl. Not a product feature. --- # Multi-Stage Dockerfile > Dockerfile-only multi-stage images **Subcategory:** GitHub **License:** MIT **Source:** https://github.com/github/awesome-copilot/blob/main/skills/multi-stage-dockerfile/SKILL.md --- ## When it is useful Create optimized multi-stage Dockerfiles for any language or framework ## The Catch Dockerfile-only. Will rewrite the image even when you asked for app code. --- # MCP Security Audit > Audit .mcp.json, not security-review **Subcategory:** GitHub **License:** MIT **Source:** https://github.com/github/awesome-copilot/blob/main/skills/mcp-security-audit/SKILL.md --- ## When it is useful Audit MCP (Model Context Protocol) server configurations for security issues. Use this skill when: reviewing .mcp.json files for security risks; checking MCP server args for hardcoded secrets or shell injection patterns; validating that MCP servers use pinned versions (not @latest); detecting unpinned dependencies in MCP server configurations; auditing which MCP servers a project registers and whether they're on an approved list; checking for environment variable usage vs. hardcoded credentials in MCP configs; any request like "is my MCP config secure?", "audit my MCP servers", or "check .mcp.json". ## The Catch Audits .mcp.json for secrets, shell injection, unpinned @latest. Config-only — not security-review. --- # OpenAPI to Application Code > Codegen a whole app from an OpenAPI spec **Subcategory:** GitHub **License:** MIT **Source:** https://github.com/github/awesome-copilot/blob/main/skills/openapi-to-application-code/SKILL.md --- ## When it is useful Generate a complete, production-ready application from an OpenAPI specification ## The Catch Loud: codegen mega-skill; will invent a whole production app from a spec. --- # Safe Browser > Local domain-allowlisted Claude Agent SDK browser **Subcategory:** Browserbase **License:** MIT (frontmatter; repo LICENSE 404) **Source:** https://github.com/browserbase/skills/blob/main/skills/safe-browser/SKILL.md --- ## When it is useful Build local constrained-browser agents with a safe_browser tool that owns CDP, enforces a domain allowlist with Fetch interception, and lets a runtime Claude Agent SDK agent complete browsing tasks without raw browser, shell, or CDP access. Use when the user wants an agent to browse or scrape while staying on approved domains, demo blocked off-domain navigation, or generate a safe browser client. ## The Catch Local domain-allowlisted browser (Claude Agent SDK). Not Browserbase cloud. Loud: undeclared-repo LICENSE. --- # Browserbase Functions > Scheduled or webhook Browserbase cloud browser jobs **Subcategory:** Browserbase **License:** MIT (frontmatter; repo LICENSE 404) **Source:** https://github.com/browserbase/skills/blob/main/skills/functions/SKILL.md --- ## When it is useful Deploy serverless browser automation as cloud functions using Browserbase. Use when the user wants to deploy browser automation to run on a schedule or cron, create a webhook endpoint for browser tasks, run automation in the cloud instead of locally, or asks about Browserbase Functions. ## The Catch Browserbase Functions — scheduled/webhook cloud browser jobs. Loud: paid product path. --- # Firecrawl Map > URL list/filter. Discovery only; not extract **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-map/SKILL.md --- ## When it is useful Discover and list a site's URLs, with search filtering. Use for "map the site" or "find the URL for" requests — when the user knows the site but not the exact page, or wants site structure. ## The Catch URL list/filter. Discovery only; not extract. --- # Firecrawl Download > Bulk local markdown and screenshots **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-download/SKILL.md --- ## When it is useful Save a site or section as local files (markdown, screenshots). Use for "download the site", offline docs, or a local copy for reference. ## The Catch bulk local markdown/screenshots. Loud: credit-heavy. --- # Firecrawl Monitor > Webhook/email watches; recurring paid path **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-monitor/SKILL.md --- ## When it is useful Alert by webhook/email on web changes — use for "monitor/watch/track/alert me when": recurring checks on known URLs (prefer over repeated one-off scrapes) or web-wide watches for new results (queries + goal). ## The Catch webhook/email watches. Loud: recurring paid path. --- # Instrument Integration > PostHog SDK wiring. Free, not a trial **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/instrument-integration/SKILL.md --- ## When it is useful Add PostHog SDK integration to your application. Use when setting up PostHog for the first time or reviewing PRs that need PostHog initialization. Covers SDK installation, provider setup, and basic configuration for any framework. ## The Catch PostHog SDK wiring. Free on posthog.com/pricing (not a trial). --- # Instrument Product Analytics > PostHog product analytics events **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/instrument-product-analytics/SKILL.md --- ## When it is useful Add PostHog product analytics events to track user behavior. Use after implementing new features or reviewing PRs to ensure meaningful user actions are captured. Also handles initial PostHog SDK setup if not yet installed. ## The Catch PostHog product analytics events. Free tier; not a trial. --- # Instrument Feature Flags > PostHog feature flags. Free tier **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/instrument-feature-flags/SKILL.md --- ## When it is useful Add PostHog feature flags to gate new functionality. Use after implementing features or reviewing PRs to ensure safe rollouts with feature flag controls. Also handles initial PostHog SDK setup if not yet installed. ## The Catch PostHog feature flags. Free tier; not a trial. --- # Instrument LLM Analytics > PostHog LLM analytics. Free tier **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/instrument-llm-analytics/SKILL.md --- ## When it is useful Add PostHog LLM analytics to trace AI model usage. Use after implementing LLM features or reviewing PRs to ensure all generations are captured with token counts, latency, and costs. Also handles initial PostHog SDK setup if not yet installed. ## The Catch PostHog LLM analytics. Free tier; not a trial. --- # Liveblocks Best Practices > Liveblocks collab. Free on liveblocks.io/pricing **Subcategory:** Liveblocks **License:** Apache-2.0 **Source:** https://github.com/liveblocks/skills/blob/main/skills/liveblocks-best-practices/SKILL.md --- ## When it is useful Use this skill when building, debugging, or answering questions about Liveblocks. Liveblocks gives you the building blocks and infrastructure to enable people and AI to work together inside your app, powering realtime collaboration. Liveblocks features include collaboration, rooms, organizations, workspaces, comments, composer, threads, notifications, multiplayer, conflict resolution, realtime presence, avatar stacks, AI collaborators, AI agents, text editors, Tiptap, BlockNote, Lexical, React Flow, Chat SDK. Common components include AiChat, Thread, InboxNotification, Composer, Toolbar (for Lexical Tiptap), FloatingToolbar, FloatingComposer, FloatingThreads, AnchoredThreads. Common hooks include useThreads, useStorage, useMutation, useOthers, useInboxNotifications, useAiChats. Common issues are related to authentication (ID tokens vs access tokens), permissions, room limits, connection errors, user info. ## The Catch Liveblocks collab. Free on liveblocks.io/pricing; no CC. --- # Yjs Best Practices > Yjs library, not Liveblocks-specific **Subcategory:** Liveblocks **License:** Apache-2.0 **Source:** https://github.com/liveblocks/skills/blob/main/skills/yjs-best-practices/SKILL.md --- ## When it is useful Use this skill when implementing or debugging Yjs features. It helps with Yjs import issues, subdocuments, `YKeyValue`, `Y.Map`, and experimental V2 encoding. ## The Catch Yjs library, not Liveblocks-specific. --- # Stripe Best Practices > Stripe account exists; processing fees apply **Subcategory:** Stripe **License:** MIT **Source:** https://github.com/stripe/ai/blob/main/skills/stripe-best-practices/SKILL.md --- ## When it is useful Guides Stripe integration decisions across API selection (Checkout Sessions vs PaymentIntents), Connect platform setup (Accounts v2, controller properties), billing/subscriptions, tax and registrations (Stripe Tax, automatic_tax, product tax codes), Treasury financial accounts, integration options (Checkout, Payment Element), migrating from deprecated Stripe APIs, and security best practices (API key management, restricted keys, webhooks, OAuth). Use when building, modifying, or reviewing any Stripe integration, including accepting payments, building marketplaces, integrating Stripe, processing payments, setting up subscriptions, collecting sales tax, VAT, or GST, creating connected accounts, or implementing secure key handling. ## The Catch Loud: Stripe account exists; processing fees apply. Toolkit MIT. --- # Stripe Docs > Stripe docs lookup. Processing fees apply **Subcategory:** Stripe **License:** MIT **Source:** https://github.com/stripe/ai/blob/main/skills/stripe-docs/SKILL.md --- ## When it is useful Use when the user or agent needs to read, search, or look up Stripe documentation or API reference. Prefer this over curl or WebFetch for any docs.stripe.com content. ## The Catch Stripe docs lookup. Loud: processing fees apply. --- # Upgrade Stripe > Stripe API/SDK upgrade path **Subcategory:** Stripe **License:** MIT **Source:** https://github.com/stripe/ai/blob/main/skills/upgrade-stripe/SKILL.md --- ## When it is useful Guide for upgrading Stripe API versions and SDKs ## The Catch Stripe API/SDK upgrade path. Loud: processing fees apply. --- # Migrate AI SDK v6 to v7 > Vercel AI SDK v6→v7. Version-gate; will stale **Subcategory:** Vercel **License:** Apache-2.0 **Source:** https://github.com/vercel/ai/blob/main/skills/migrate-ai-sdk-v6-to-v7/SKILL.md --- ## When it is useful Migrate applications from AI SDK 6.x to AI SDK 7.0. Use when upgrading Vercel AI SDK packages, fixing v7 migration errors, or when the user mentions AI SDK v6, v7, upgrade, migration, breaking changes, system to instructions, fullStream, telemetry, tool context, or finalStep. ## The Catch Vercel AI SDK v6→v7. Version-gate; will stale. --- # Add WebMCP > First-party WebMCP; not generic MCP builder **Subcategory:** Browserbase **License:** MIT (frontmatter; no LICENSE file) **Source:** https://github.com/browserbase/skills/blob/main/skills/add-webmcp/SKILL.md --- ## When it is useful Analyze an existing web application, identify safe user-visible capabilities across routes, forms, server actions, handlers, and schemas, then implement first-party WebMCP tools and validate discovery and invocation with Stagehand. Use when the user asks to make a codebase agent-ready, expose website features as WebMCP tools, or add WebMCP directly to an app rather than generating a standalone injection script from a URL. ## The Catch first-party WebMCP; not generic MCP builder. Loud: undeclared-repo LICENSE (frontmatter MIT; repo LICENSE 404). --- # Cookie Sync > Local Chrome cookies into Browserbase **Subcategory:** Browserbase **License:** MIT (frontmatter; no LICENSE file) **Source:** https://github.com/browserbase/skills/blob/main/skills/cookie-sync/SKILL.md --- ## When it is useful Sync cookies from local Chrome to a Browserbase persistent context so the browse CLI can access authenticated sites. Use when the user wants to browse as themselves, sync cookies, or log into sites via Browserbase. ## The Catch local Chrome cookies into Browserbase; never-fires without Browserbase. Loud LICENSE (frontmatter MIT; repo LICENSE 404). --- # Neon Postgres > Lakebase Postgres setup, pooled vs direct, DATABASE_URL **Subcategory:** Neon **License:** Apache-2.0 **Source:** https://github.com/neondatabase/agent-skills/blob/main/skills/neon-postgres/SKILL.md --- ## When it is useful Guides and best practices for Lakebase Postgres on Neon: setup, connection methods and drivers, pooled vs direct connections, DATABASE_URL, branching, schema migrations, autoscaling, scale-to-zero, instant restore, read replicas, connection pooling, IP allow lists, and logical replication. Also Lakebase Search (semantic, BM25, hybrid). Use when users ask about Lakebase Postgres, Neon setup, connect to Neon, Neon project, DATABASE_URL, serverless Postgres, Neon CLI, Neon MCP, Neon Auth, @neondatabase/serverless, @neondatabase/neon-js, scale to zero, Neon autoscaling, Neon read replica, Neon connection pooling, schema migrations, semantic search, vector search, full-text search, BM25, or hybrid search. ## The Catch Vendor lock: Neon/Lakebase setup, pooled vs direct, DATABASE_URL. Trigger steals Prisma/Supabase Postgres work. Free $0 at neon.tech/pricing (100 projects, 0.5 GB, 100 CU-hours; compute suspends at cap). Parent neon stays OUT (HOLD — always-on overview). --- # Neon Postgres Branches > Pick/create Neon branch type; schema-only is Beta **Subcategory:** Neon **License:** Apache-2.0 **Source:** https://github.com/neondatabase/agent-skills/blob/main/skills/neon-postgres-branches/SKILL.md --- ## When it is useful Choose and create the right Neon branch type for testing and development. Use when users ask about Neon branching, migration testing with real data, isolated test environments, schema-only branch workflows for sensitive data, resetting a branch from its parent, branch expiration and CI/CD branch lifecycles, or branch creation via Neon CLI or Neon MCP. Triggers include "Neon branch", "test migrations safely", "branch production data", "schema-only branch", "reset branch", "branch per PR", and "sensitive data testing". Not the neon-postgres setup skill. ## The Catch Branch-type picker, not setup — distinct from neon-postgres. Schema-only branches are Beta. Free cap is 10 branches/project on neon.tech/pricing. --- # Neon Functions > Long-running Node HTTP on a Neon branch. Public beta **Subcategory:** Neon **License:** Apache-2.0 **Source:** https://github.com/neondatabase/agent-skills/blob/main/skills/neon-functions/SKILL.md --- ## When it is useful Long-running, serverless Node.js HTTP functions deployed onto a Neon branch, with DATABASE_URL injected automatically and compute next to the data. Use when a user wants to host an API, an AI agent with long streaming responses, a WebSocket or SSE server, a webhook handler, a Discord bot, an MCP server, or any request/response workload that risks timing out on short lambda-style serverless — and wants it to branch with their database. Triggers include "Neon Functions", "Neon Compute", "function that won't time out", "run code next to my database", "function logs". Do not use for Browserbase Functions or generic AWS/Vercel lambda work that is not Neon. ## The Catch Public beta. Vendor lock. False-fires on generic serverless — this is Neon Functions, not Vercel/AWS. NAME-COLLIDES with published Browserbase functions; slug is neon-functions, never functions. Free includes Functions capacity; costs past limits. --- # Neon Object Storage > S3-compatible storage that branches with Neon. Public beta **Subcategory:** Neon **License:** Apache-2.0 **Source:** https://github.com/neondatabase/agent-skills/blob/main/skills/neon-object-storage/SKILL.md --- ## When it is useful S3-compatible object storage that branches with a Neon project so files and database rows stay in sync across dev, preview, and production. Use when the user wants buckets that fork with a Neon branch, uploads or attachments keyed from Postgres rows on the same branch, or S3 SDK/CLI access against Neon Object Storage. Not for static site hosting, AWS S3, Cloudflare R2, or UploadThing. ## The Catch Public beta. Not AWS S3 and not R2 — Neon-only buckets that branch with Postgres. Vendor lock. Costs past the free Object Storage allowance. --- # LiveKit Agents > Voice AI agent via LiveKit Cloud + Agents SDK **Subcategory:** LiveKit **License:** MIT **Source:** https://github.com/livekit/agent-skills/blob/main/skills/livekit-agents/SKILL.md --- ## When it is useful Build voice AI agents with LiveKit Cloud and the Agents SDK. Use when the user asks to "build a voice agent", "create a LiveKit agent", "add voice AI", "implement handoffs", "structure agent workflows", or is working with LiveKit Agents SDK. Opinionated path is LiveKit Cloud + LiveKit Inference. Requires writing tests for all implementations. Skip when the job is self-hosting the LiveKit SFU or the deferred livekit-simulations skill. ## The Catch Cloud-only skill: skips self-host OSS. LiveKit Cloud + Inference is the path. Free Build $0, no CC, at livekit.com/pricing; Inference credits then paid. livekit-simulations stays DEFER. --- # Agent OWASP Compliance > OWASP ASI Top 10 audit before prod. Community Copilot skill **Subcategory:** GitHub **License:** MIT **Source:** https://github.com/github/awesome-copilot/blob/main/skills/agent-owasp-compliance/SKILL.md --- ## When it is useful Check any AI agent codebase against the OWASP Agentic Security Initiative (ASI) Top 10 risks. Use when evaluating an agent system's security posture before production deployment; running a compliance check against OWASP ASI 2026 standards; mapping existing security controls to the 10 agentic risks; generating a compliance report for security review or audit; comparing agent framework security features against the standard; or any request like "is my agent OWASP compliant?", "check ASI compliance", or "agentic security audit". Not for .mcp.json config audits (mcp-security-audit) and not Sentry security-review. ## The Catch Community Copilot skill, not OWASP-official. ASI Top 10 agent audit before prod — distinct from mcp-security-audit (.mcp.json) and from Sentry security-review. --- # Inngest Setup > Inngest Cloud + TypeScript setup. Hobby $0/mo, no CC **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-setup/SKILL.md --- ## When it is useful Use when adding durable execution to a TypeScript project — retry-safe webhook handlers, background jobs that survive crashes, scheduled tasks, or long-running workflows that outlive a single request. Covers Inngest SDK installation, client config, environment variables, serve endpoints (Next.js, Express, Hono, Fastify), connect-as-worker mode, and the local dev server. TypeScript-focused; Python/Go belong in Inngest docs, not this skill. ## The Catch Assumes Inngest Cloud + TypeScript. Hobby is $0/mo, no CC, at inngest.com/pricing (50k executions, 5 concurrent steps, 500k events; pauses at quota). --- # Inngest Durable Functions > Inngest steps, not BullMQ. Cloud + TypeScript **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-durable-functions/SKILL.md --- ## When it is useful Use when building functions that must survive process crashes, retry automatically on failure, run on a schedule, react to events, or maintain state across infrastructure failures — webhook handlers that drop events, flaky cron jobs, background jobs that fail mid-execution, or workflows that need to resume. Covers Inngest function configuration, triggers (events, cron, invoke), step execution and memoization, idempotency, cancellation, error handling, retries, logging, and observability. Not BullMQ, not Trigger.dev, not generic queues. ## The Catch False-fires on BullMQ (and generic queues). This is Inngest durable functions — Cloud + TypeScript. Hobby pauses at quota. --- # Inngest Events > Inngest event schema; 24h dedupe is vendor-specific **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-events/SKILL.md --- ## When it is useful Use when designing event-driven workflows, decoupling services, implementing fan-out (one trigger, many downstream handlers), implementing idempotent event handling with IDs, or handling at-least-once delivery from external sources like Stripe webhooks. Covers Inngest event schema, payload format, naming conventions, IDs for idempotency, the ts param, fan-out patterns, and system events like inngest/function.failed. ## The Catch 24-hour event-ID dedupe is Inngest-specific — not Kafka, not SQS, not a generic idempotency key. Cloud + TypeScript. --- # Inngest Flow Control > Concurrency/throttle/batch. Hobby is 5 concurrent steps **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-flow-control/SKILL.md --- ## When it is useful Use when handling external API rate limits (OpenAI 429s, HubSpot or Stripe), preventing duplicate work from rapid event bursts (debouncing user actions), spreading load over time, ensuring per-tenant fairness, processing events in batches, limiting concurrent runs of the same operation, or assigning priority to important runs. Covers Inngest flow control: concurrency limits with keys, throttling, rate limiting, debounce, priority, singleton, and event batching. ## The Catch Loud: Hobby is 5 concurrent steps. This skill will prescribe concurrency, throttle, and batching the free plan cannot run. Upgrade or the controls lie. --- # Inngest Realtime > v4-native inngest/realtime, not @inngest/realtime **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-realtime/SKILL.md --- ## When it is useful Use when streaming durable workflow updates to a UI in real time — live order status, AI agent token streaming from a function to the browser, log tailing for long-running jobs, or human-in-the-loop approval flows. Covers Inngest v4 native realtime: typed channels, publishing from inside step.run, minting subscription tokens via server actions, and consuming the stream from React/Next.js. Do not install the v3-era @inngest/realtime package on inngest@4. ## The Catch v4-native: import from inngest/realtime. @inngest/realtime is the v3 package and will 401/TypeError on v4. Cloud + TypeScript. --- # Inngest Agents > AgentKit lock. Durable agents on Inngest Cloud **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-agents/SKILL.md --- ## When it is useful Use when building durable AI agents or agentic workflows with Inngest and AgentKit, including model calls, tool calls, multi-agent networks, human approval, realtime progress, provider rate limits, crash-safe execution, and Agent Evals handoff. Covers AgentKit, step.ai, step.run, step.waitForEvent, native realtime, and when to use lower-level Inngest primitives instead of an in-memory agent loop. Skip generic LangGraph/CrewAI work that is not Inngest AgentKit. ## The Catch AgentKit lock: createAgent + step.ai on Inngest Cloud. Not a generic agent framework skill. Hobby 5-step cap still applies. --- # Inngest v3 to v4 Migration > SDK v3→v4. Stale the day v5 ships **Subcategory:** Inngest **License:** Apache-2.0 **Source:** https://github.com/inngest/inngest-skills/blob/main/skills/inngest-v3-v4-migration/SKILL.md --- ## When it is useful Use when upgrading an existing TypeScript codebase from Inngest SDK v3 to v4, or when fixing mixed v3/v4 API usage. Covers detecting current SDK usage, moving triggers into createFunction options, replacing EventSchemas with eventType/staticSchema, moving serve options to the client, updating realtime imports, rewriting step.invoke string IDs, checkpointing/serverless runtime settings, Connect option changes, and verification. ## The Catch Version-gate: v3→v4 only. Stale the day v5 ships. Not a greenfield setup skill. --- # Convex Quickstart > Next.js + Convex template. JSON-generated; can drift **Subcategory:** Convex **License:** Apache-2.0 **Source:** https://github.com/get-convex/agent-skills/blob/main/skills/convex-quickstart/SKILL.md --- ## When it is useful Get a barebones Convex + web template running from a one-sentence idea. Stand up a Next.js + Convex template locally with an anonymous dev deployment. Minimal: local dev servers, no publish step, no pre-baked auth. Use when starting a new Convex project or adding Convex to an existing app. Parent convex is HOLD — do not treat this as the always-on Convex overview. ## The Catch Next.js template path. SKILL.md is JSON-generated and can drift from live Convex APIs. Loud: convex.dev/pricing Free has hard caps; Starter is $0 base with a card. Parent convex stays HOLD. --- # Convex Auth > Convex Auth only — not Clerk or Better Auth **Subcategory:** Convex **License:** Apache-2.0 **Source:** https://github.com/get-convex/agent-skills/blob/main/skills/convex-auth/SKILL.md --- ## When it is useful Add authentication (passkeys/OAuth) to the current Convex app, including auth.config.ts wiring. Installs and wires @convex-dev/auth: provider, server config, client hooks, and a sign-in UI. Passkeys by default; password/OAuth only on explicit request. Not Clerk, not Better Auth, not Auth.js, not WorkOS. ## The Catch Convex Auth (@convex-dev/auth) only — not Clerk, not Better Auth. Missing auth.config.ts silently always-signed-out. Loud Free/Starter card on convex.dev/pricing. --- # Convex Create Component > Convex Components with isolated tables, not React **Subcategory:** Convex **License:** Apache-2.0 **Source:** https://github.com/get-convex/agent-skills/blob/main/skills/convex-create-component/SKILL.md --- ## When it is useful Build reusable Convex components with isolated tables and a small app-facing API. Use for new Convex components, reusable backend modules, integrations, or component boundary work. Not for one-off business logic, thin TypeScript utilities, React components, or app-level orchestration that should stay in convex/. ## The Catch Convex Components — backend modules with isolated tables — not React components. Loud Free/Starter card. --- # Convex Migrate > Live-data backfill with @convex-dev/migrations, not Prisma **Subcategory:** Convex **License:** Apache-2.0 **Source:** https://github.com/get-convex/agent-skills/blob/main/skills/convex-migrate/SKILL.md --- ## When it is useful Migrate schema and backfill data on a deployed Convex app using @convex-dev/migrations. Stage the schema change (new fields optional first), write a backfill that makes old rows valid, run it, verify, then tighten the validator. Use for live Convex data migrations. Not Prisma, not Drizzle, not neon-postgres-branches. ## The Catch Live-data backfill on Convex, not Prisma. Tightening a validator before backfill rejects existing rows. Loud Free/Starter card. --- # Infisical Setup > CLI/SDK secrets into the app. Slug is infisical-setup **Subcategory:** Infisical **License:** MIT **Source:** https://github.com/Infisical/ai-skills/blob/main/skills/infisical-setup/SKILL.md --- ## When it is useful Interactive setup for using Infisical as a secret management tool: local development (CLI), Docker (build-time and runtime injection), CI/CD (GitHub Actions, GitLab CI), Kubernetes (Operator + CRDs belong in infisical-kubernetes-operator), and application code (Node.js, Python, Go, Java, .NET, Ruby, PHP, Rust, C++ SDKs). Also machine identity auth methods. Use when someone asks about using Infisical, injecting secrets, infisical run, infisical init, Docker secrets, SDK initialization, or CI/CD secret injection. Not for pushing secrets out (infisical-secret-syncs) or self-hosting the platform (infisical-self-host). ## The Catch Folder is infisical-setup; frontmatter name is infisical-user-setup-guide — catalog slug is infisical-setup. Cloud Free $0 forever (5 identities, unlimited projects, 3 envs, 10 secret syncs) at infisical.com/pricing. Not the k8s operator CRDs. --- # Infisical Kubernetes Operator > v1beta1 CRDs. v1alpha1 is a false-fire **Subcategory:** Infisical **License:** MIT **Source:** https://github.com/Infisical/ai-skills/blob/main/skills/infisical-kubernetes-operator/SKILL.md --- ## When it is useful Use when installing or operating the Infisical Kubernetes Operator: Helm install, InfisicalSecret / InfisicalStaticSecret CRDs, auto-reload, and pushing secrets into cluster resources. Covers v1beta1 CRDs and legacy v1alpha1. Not generic External Secrets Operator, not infisical-setup SDK wiring, not infisical-self-host of the platform. ## The Catch v1alpha1 is a false-fire — current CRDs are v1beta1. Dynamic-secret leases are paid Advanced even on self-host. Not the setup skill. --- # Infisical Self-Host > MIT core only. ee/ needs LICENSE_KEY **Subcategory:** Infisical **License:** MIT **Source:** https://github.com/Infisical/ai-skills/blob/main/skills/infisical-self-host/SKILL.md --- ## When it is useful Deploy and operate Infisical self-hosted instances with Docker, Docker Compose, and Kubernetes. Covers architecture, environment variables, ENCRYPTION_KEY management, PostgreSQL, Redis (required noeviction), production hardening, FIPS, scaling, and HA. For deploying the Infisical platform itself. Not the Kubernetes Operator (separate Helm chart) and not using Infisical once running (infisical-setup). ## The Catch MIT core only. ee/ is not MIT — needs LICENSE_KEY. This deploys the platform; it does not inject secrets into an app. --- # Infisical Secret Scanning > github-radar data source, not github. infisical scan **Subcategory:** Infisical **License:** MIT **Source:** https://github.com/Infisical/ai-skills/blob/main/skills/infisical-secret-scanning/SKILL.md --- ## When it is useful Use when scanning repos and pipelines for leaked secrets with Infisical: GitHub/GitLab/Bitbucket data sources, infisical scan CLI, pre-commit hooks, and honey tokens. Not generic gitleaks-only work unless Infisical scanning is the product, and not infisical-setup secret injection. ## The Catch GitHub data source is github-radar, not github. Distinct from setup and from syncs. Cloud Free $0 forever still caps identities and syncs. --- # Infisical Secret Syncs > Push secrets out. Free cap is 10 syncs **Subcategory:** Infisical **License:** MIT **Source:** https://github.com/Infisical/ai-skills/blob/main/skills/infisical-secret-syncs/SKILL.md --- ## When it is useful Use when pushing Infisical secrets outward to third-party destinations (the 48 sync destinations, key schemas, initial-sync enums). This is secrets leaving Infisical. Not injecting secrets into an app (infisical-setup), not dynamic-secret leases, not secret scanning. ## The Catch Push out, not pull in. Cloud Free cap is 10 secret syncs. Distinct from infisical-setup. --- # React Email > Compose React email templates. Delivery still hits 100/day **Subcategory:** Resend **License:** MIT **Source:** https://github.com/resend/resend-skills/blob/main/skills/react-email/SKILL.md --- ## When it is useful Use when building HTML email templates with React components, adding a visual email editor, rendering emails to HTML, or composing welcome emails, password resets, notifications, order confirmations, newsletters, and transactional templates. This is compose, not the HOLD parent resend send-API skill. Delivery through Resend still hits the Free 100/day and 3,000/mo caps. ## The Catch Compose templates. Sending still hits Resend Free $0/mo (3,000 emails/mo, 100/day, no overage). Overlaps HOLD parent resend on “sending”. Distinct from resend-cli. --- # Resend CLI > resend CLI for shell/CI. --react-email can steal compose **Subcategory:** Resend **License:** MIT **Source:** https://github.com/resend/resend-skills/blob/main/skills/resend-cli/SKILL.md --- ## When it is useful Operate Resend from the terminal — send emails (including React Email .tsx via --react-email), manage domains, contacts, broadcasts, templates, webhooks, API keys, logs, automations, and events via the resend CLI. Use when the user wants Resend commands in the shell, scripts, or CI/CD. Always load before running resend commands. Parent resend and agent-email-inbox stay HOLD. ## The Catch Shell/CI. --react-email can steal react-email compose work. Free $0/mo still 3,000/mo and 100/day with no overage — the CLI will fail at the cap, not bill. --- # Drizzle Migration Conflict > Community v0.1. Not drizzle-team. Can steal generate **Subcategory:** Other **License:** Apache-2.0 **Source:** https://github.com/chaunsin/agent-skills/blob/master/skills/drizzle-migration-conflict/SKILL.md --- ## When it is useful Diagnose, repair, and prevent Drizzle Kit migration conflicts in team workflows: _journal.json, snapshot.json, non-commutative generated SQL, merge/rebase conflicts, drizzle-kit check. Prefer this when a Drizzle migration file conflicts after a pull, merge, rebase, or PR update. Official drizzle-team kit skills 404 on main. Not a clean-greenfield drizzle-kit generate unless a conflict already exists. hugo-to-markdown stays DEFER. ## The Catch Community v0.1, not drizzle-team. Official kit skills 404 on main. Can steal a clean drizzle-kit generate. --- # ElevenLabs Text-to-Speech > Cloud TTS. Free 10k credits is minutes, not a product **Subcategory:** ElevenLabs **License:** MIT **Source:** https://github.com/elevenlabs/skills/blob/main/text-to-speech/SKILL.md --- ## When it is useful ElevenLabs cloud TTS. Needs API key. Not SFX, music, dubbing, Speech Engine, or LiveKit OSS. ## The Catch Vendor lock. Needs API key. Free 10k credits/mo is minutes of audio, not a product. No commercial license on Free. Starter is 6 USD/mo not 5. --- # ElevenLabs Speech Engine > Your runtime, their STT/TTS. Cloud + key. Not LiveKit OSS **Subcategory:** ElevenLabs **License:** MIT **Source:** https://github.com/elevenlabs/skills/blob/main/speech-engine/SKILL.md --- ## When it is useful Real-time voice on a custom agent runtime via ElevenLabs Speech Engine (WebSocket, WebRTC, conversation tokens). They handle mic/STT/turn-taking/TTS; your server owns agent logic. Needs API key. Not LiveKit Agents OSS. Not the TTS-only skill. ## The Catch Your runtime, their STT/TTS. Cloud + API key, not LiveKit OSS. Free has no commercial license. Distinct from elevenlabs-text-to-speech. --- # ElevenLabs Sound Effects > Text-to-SFX, not speech. 200 credits/gen **Subcategory:** ElevenLabs **License:** MIT **Source:** https://github.com/elevenlabs/skills/blob/main/sound-effects/SKILL.md --- ## When it is useful Generate sound effects from text with ElevenLabs: ambient, cinematic, UI, looping, duration. Not speech, not music, not dubbing. Needs API key. JS: scoped elevenlabs-js. ## The Catch Distinct from TTS. 200 credits/gen. Free 10k credits/mo; no commercial license on Free. Needs API key. --- # ElevenLabs Music > Music API. Commercial music is Starter+. 900 credits/min **Subcategory:** ElevenLabs **License:** MIT **Source:** https://github.com/elevenlabs/skills/blob/main/music/SKILL.md --- ## When it is useful Generate music with the ElevenLabs Music API: instrumentals, lyrics, jingles, composition plans. Not TTS, SFX, or dubbing. Needs API key. ## The Catch Commercial music is Starter+ (6 USD/mo, not 5). 900 credits/min. Free has no commercial license. Distinct from TTS and sound-effects. --- # ElevenLabs Dubbing > Dubbing Studio is Starter+. 2k-10k credits/min **Subcategory:** ElevenLabs **License:** MIT **Source:** https://github.com/elevenlabs/skills/blob/main/dubbing/SKILL.md --- ## When it is useful Dub audio or video into other languages with ElevenLabs Dubbing API v2, preserving original speakers. Use Dubbing Projects API, not legacy v1 create. Not TTS, SFX, or music. Needs API key. ## The Catch Dubbing Studio is Starter+ (6 USD/mo). 2k-10k credits/min. Free has no commercial license. Distinct from TTS. Needs API key. --- # Replicate Find Models > Search Replicate models. Not Hugging Face Hub **Subcategory:** Replicate **License:** Apache-2.0 **Source:** https://github.com/replicate/skills/blob/main/skills/find-models/SKILL.md --- ## When it is useful Find AI models on Replicate via search and curated collections. Always query the Replicate API for current models. Not Hugging Face Hub search. Not compare-models (cost/speed/quality pick) and not run-models. ## The Catch Vendor search on Replicate, not Hugging Face. No ongoing 0 USD hobby. PAYG at replicate.com/pricing. Try-for-Free is a one-time limited run, not a plan. --- # Replicate Compare Models > Pick among Replicate models. Cost numbers stale **Subcategory:** Replicate **License:** Apache-2.0 **Source:** https://github.com/replicate/skills/blob/main/skills/compare-models/SKILL.md --- ## When it is useful Compare Replicate models by cost, speed, quality, and capabilities after a shortlist from find-models. Not discovery search. Cost and latency figures will go stale. ## The Catch Pick vs find. Cost numbers will stale. No ongoing 0 USD hobby. PAYG. Try-for-Free is not a plan. --- # Replicate Run Models > Paid predictions. Needs REPLICATE_API_TOKEN **Subcategory:** Replicate **License:** Apache-2.0 **Source:** https://github.com/replicate/skills/blob/main/skills/run-models/SKILL.md --- ## When it is useful Run Replicate models via predictions, webhooks, streaming, files, concurrency. Needs REPLICATE_API_TOKEN. This is the paid API path. Try-for-Free UI is not this skill. ## The Catch Paid path. Needs REPLICATE_API_TOKEN. Try-for-Free is a one-time limited run, not this skill and not a plan. PAYG at replicate.com/pricing. --- # Replicate Build Models > Cog package. Skill still aims at Replicate deploy **Subcategory:** Replicate **License:** Apache-2.0 **Source:** https://github.com/replicate/skills/blob/main/skills/build-models/SKILL.md --- ## When it is useful Package custom models with Cog (cog.yaml, predict.py, local serve/predict, GPU/Docker). Still aims at Replicate deploy. For pushing see publish-models. ## The Catch Cog is OSS; skill still aims at Replicate deploy. Local GPU/Docker. No 0 USD hobby. Try-for-Free is not a plan. Distinct from run and publish. --- # Replicate Publish Models > cog push. Public model is not free inference **Subcategory:** Replicate **License:** Apache-2.0 **Source:** https://github.com/replicate/skills/blob/main/skills/publish-models/SKILL.md --- ## When it is useful Push custom models to Replicate with cog push or cog-safe-push and CI/CD. Not packaging (build-models) and not running others models (run-models). ## The Catch cog push. Public model is not free inference. PAYG, no 0 USD hobby. Try-for-Free is a one-time limited run, not a plan. --- # Sanity Migration > Replatform to Sanity. Full CMS move, not PT conversion **Subcategory:** Sanity **License:** MIT **Source:** https://github.com/sanity-io/agent-toolkit/blob/main/skills/sanity-migration/SKILL.md --- ## When it is useful Plans CMS-to-Sanity replatforms from AEM, Contentful, Strapi, Webflow, WordPress, Payload, Drupal, Markdown/MDX. Full content strategy and ETL, not HTML/MD to Portable Text ingest alone, and not PT render. ## The Catch Vendor lock. Full CMS move, not PT conversion. Free 0 USD forever (20 seats, 2 permission roles, 2 datasets public only). Growth 15 USD/seat. --- # Sanity Portable Text Conversion > HTML/MD to Portable Text. Ingest, not replatform **Subcategory:** Sanity **License:** MIT **Source:** https://github.com/sanity-io/agent-toolkit/blob/main/skills/portable-text-conversion/SKILL.md --- ## When it is useful Convert HTML and Markdown into Portable Text for Sanity with @portabletext/markdown and @portabletext/block-tools. Ingest pipelines, not a full CMS replatform (sanity-migration) and not rendering PT out (serialization). ## The Catch Ingest, not replatform. Distinct from sanity-migration and from serialization (render). Free 0 USD forever still public-only datasets (2) and 2 roles. Growth 15 USD/seat. --- # Sanity Portable Text Serialization > PT to React/Svelte/Vue/HTML. Render, not ingest **Subcategory:** Sanity **License:** MIT **Source:** https://github.com/sanity-io/agent-toolkit/blob/main/skills/portable-text-serialization/SKILL.md --- ## When it is useful Render Portable Text to React, Svelte, Vue, Astro, HTML, Markdown, plain text with @portabletext libraries. Custom serializers. Not converting HTML/MD into PT (conversion) and not CMS replatform (migration). ## The Catch Render, not ingest. Distinct from conversion and from sanity-migration. Free 0 USD forever: 20 seats, 2 roles, 2 public datasets. Growth 15 USD/seat. --- # TanStack AI Chat Experience > False-fires on Vercel AI SDK streamText / useChat **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/chat-experience/SKILL.md --- ## When it is useful Build streaming chat with TanStack AI, not Vercel AI SDK streamText or useChat from ai. Sample path uses TanStack Start. Pinned library 0.42.0. OSS SDK is 0 USD; model calls need provider keys. ## The Catch False-fires on Vercel AI SDK streamText / useChat from ai. Distinct from migrate-ai-sdk-v6-to-v7. Pinned 0.42.0. OSS SDK 0 USD; inference still needs provider keys. --- # TanStack AI Tool Calling > Isomorphic toolDefinition. Not OpenAI tools / Vercel tool() **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/tool-calling/SKILL.md --- ## When it is useful Type-safe isomorphic tools via TanStack AI toolDefinition that can run on server or client. Not OpenAI tools arrays and not Vercel AI SDK tool(). Pinned 0.42.0. ## The Catch Isomorphic toolDefinition. False-fires on OpenAI tools and Vercel tool(). Distinct from chat-experience. OSS SDK 0 USD; provider keys still required. --- # TanStack AI Media Generation > Not ElevenLabs/Replicate. generate* needs provider keys **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/media-generation/SKILL.md --- ## When it is useful Generate image/audio/video via TanStack AI generate helpers and adapters. Not ElevenLabs skills and not Replicate run-models. Every generate call needs a paid provider key. Pinned 0.42.0. ## The Catch Not ElevenLabs or Replicate. Every generate-star needs a paid provider key. OSS SDK is 0 USD. Distinct from chat and tool-calling. Pinned 0.42.0. --- # TanStack AI Structured Outputs > Forbids provider-level json_schema you already set **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/structured-outputs/SKILL.md --- ## When it is useful Structured outputs with TanStack AI via JSON Schema, Zod, ArkType, or Valibot. Do not also set provider-level json_schema when the SDK already owns the schema. Pinned 0.42.0. Not Vercel AI SDK generateObject. ## The Catch Forbids provider-level json_schema you already set on the SDK. Distinct from tool-calling. OSS 0 USD; provider keys still required. Pinned 0.42.0. --- # TanStack AI Adapter Configuration > BytePlus TWO keys no fallback. Ollama is the 0 USD local path **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/adapter-configuration/SKILL.md --- ## When it is useful Configure TanStack AI provider adapters. BytePlus needs TWO keys with no fallback. Ollama is the only 0 USD local path. Other adapters need paid provider keys. Not chat UI. Pinned 0.42.0. ## The Catch BytePlus TWO keys, no fallback. Ollama is the only 0 USD local path. Distinct from media-generation. Pinned 0.42.0. --- # TanStack AI AG-UI Protocol > Server AG-UI SSE/NDJSON. Not CopilotKit client **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-core/ag-ui-protocol/SKILL.md --- ## When it is useful Expose TanStack AI as a server AG-UI protocol over SSE or NDJSON. Not CopilotKit client widgets. Not chat-experience UI. Pinned 0.42.0. ## The Catch Server AG-UI SSE/NDJSON, not CopilotKit client. Distinct from chat-experience. OSS SDK 0 USD; provider keys still required. Pinned 0.42.0. --- # TanStack AI Code Mode > Isolate drivers. 0.3.8 vs ai-core 0.42.0. Not generic sandbox **Subcategory:** TanStack **License:** MIT **Source:** https://github.com/TanStack/ai/blob/main/skills/ai-code-mode/SKILL.md --- ## When it is useful LLM-generated TypeScript in sandboxed isolate drivers via createCodeModeTool. Package @tanstack/ai-code-mode is 0.3.8, a different release train from ai-core 0.42.0. Not a generic sandbox skill. HOLD ai-core, middleware, custom-backend-integration. ## The Catch Isolate drivers. 0.3.8 vs ai-core 0.42.0 different release train. Not a generic sandbox. Distinct from chat and tool-calling. --- # Tavily Search > LLM-optimized search. Generic steal. 1 credit basic / 2 advanced **Subcategory:** Tavily **License:** MIT **Source:** https://github.com/tavily-ai/skills/blob/main/skills/tavily-search/SKILL.md --- ## When it is useful Search the web with LLM-optimized results via the Tavily CLI (tvly). Use for search for / look up / find articles. Needs tvly login. Do not fall back to other search tools. Twin of firecrawl-build-search (that is Firecrawl /search in product code). ## The Catch Loud generic steal (search for / look up). Body says do not fall back to other tools. Twins live firecrawl-build-search. 1 credit basic / 2 advanced. Researcher Free 0 USD/mo is 1,000 credits, no CC. PAYG 0.008 USD/credit. Needs tvly CLI + login. --- # Tavily Research > Cited multi-source report. Loud free-tier eater **Subcategory:** Tavily **License:** MIT **Source:** https://github.com/tavily-ai/skills/blob/main/skills/tavily-research/SKILL.md --- ## When it is useful Cited multi-source research report via Tavily CLI. Not tavily-search fact-finding. Needs tvly login. Mini 4-110 credits, pro 15-250 vs Researcher Free 1,000/mo. ## The Catch Loud free-tier eater: mini 4-110 credits, pro 15-250 vs 1,000/mo. One pro run can burn a quarter of Researcher. Distinct from tavily-search. PAYG 0.008 USD/credit after free. --- # Pinecone Quickstart > First-time Database vs Assistant tour. Not production schema **Subcategory:** Pinecone **License:** MIT **Source:** https://github.com/pinecone-io/skills/blob/main/skills/pinecone-quickstart/SKILL.md --- ## When it is useful First-time Pinecone tour: Database (integrated index, upsert, query via MCP + Python) or Assistant. Needs PINECONE_API_KEY and MCP. Not production schema design. Not upstash-vector-js. ## The Catch First-time Database vs Assistant tour. PINECONE_API_KEY + MCP. Not production schema. Distinct from upstash-vector-js. Starter Free: 2 GB, 2M WU/mo, 1M RU/mo, AWS us-east-1 only. Builder 20 USD/mo. Standard 50 USD/mo min. --- # Pinecone Query > Integrated indexes only. Standard/external must use pinecone-cli **Subcategory:** Pinecone **License:** MIT **Source:** https://github.com/pinecone-io/skills/blob/main/skills/pinecone-query/SKILL.md --- ## When it is useful Query Pinecone integrated indexes (built-in embeddings) via MCP with natural language. Standard or external-embedding indexes must use pinecone-cli. MCP-only. Needs PINECONE_API_KEY. Not upstash-vector-js. ## The Catch LOUD: integrated indexes only (built-in embeddings). Standard/external must use pinecone-cli. MCP-only. Distinct from upstash-vector-js and from pinecone-quickstart. --- # Pinecone CLI > pc CLI for all index types. False-fires on generic pc/redis-cli **Subcategory:** Pinecone **License:** MIT **Source:** https://github.com/pinecone-io/skills/blob/main/skills/pinecone-cli/SKILL.md --- ## When it is useful Use the Pinecone pc CLI for all index types (standard, integrated, sparse), batch ops, backups, namespaces, CI/CD. Unlike MCP which is integrated-indexes only. False-fires on generic pc or redis-cli already in PATH. ## The Catch pc CLI for all index types. False-fires on generic pc/redis-cli in PATH. Distinct from pinecone-query (MCP integrated-only). Starter Free caps still apply. --- # Pinecone Full-Text Search > SDK 10+ / API 2026-07. pinecone.preview deleted. Not Elasticsearch **Subcategory:** Pinecone **License:** MIT **Source:** https://github.com/pinecone-io/skills/blob/main/skills/pinecone-full-text-search/SKILL.md --- ## When it is useful Pinecone full-text-search document index via graduated document-schema API. Python SDK >= 10.0.0, API 2026-07. pinecone.preview is deleted. Not Elasticsearch mapping. Distinct from pinecone-query vectors and from algolia search. ## The Catch SDK >=10 / API 2026-07. pinecone.preview deleted. Not Elasticsearch. Distinct from upstash-vector-js and from pinecone-query. Do not claim 5 indexes (unverified). --- # Pinecone Assistant > Managed RAG. Starter caps. Not LangChain RAG **Subcategory:** Pinecone **License:** MIT **Source:** https://github.com/pinecone-io/skills/blob/main/skills/pinecone-assistant/SKILL.md --- ## When it is useful Create and chat with Pinecone Assistant, a managed RAG service with citations. Upload docs, ask questions. Not LangChain RAG pipelines and not pinecone-query vector search. Starter Assistant caps: 1GB / 500k in / 300k out / 1k ingest. ## The Catch Managed RAG. Starter Assistant caps 1GB / 500k in / 300k out / 1k ingest. Not LangChain RAG. Distinct from pinecone-quickstart Assistant tour and from upstash-vector-js. --- # Qdrant Deployment Options > Docker vs Cloud vs local vs Edge. Local Python in-memory is not server-format **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-deployment-options/SKILL.md --- ## When it is useful Choose Qdrant deployment: Docker, Cloud, local, or Edge. Local Python in-memory is NOT server-format compatible. Generic host steal. OSS self-host is 0 USD. Not Pinecone Cloud. ## The Catch Docker vs Cloud vs local vs Edge. Local Python in-memory is NOT server-format compatible. Generic host steal. Cloud Free unused clusters suspended after 1 week, deleted after 4 weeks. Distinct from pinecone-quickstart. --- # Qdrant Edge > In-process shard. Beta API drifts. No one-call .sync(). Not sqlite-vec **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-edge/SKILL.md --- ## When it is useful Build on Qdrant Edge, the embedded in-process shard (Python or Rust). Beta API drifts. No one-call .sync() to Cloud. Not sqlite-vec. Distinct from qdrant-deployment-options. ## The Catch In-process shard. Beta API drifts. No one-call .sync(). Not sqlite-vec. Distinct from deployment-options. Cloud Free unused clusters still suspend after 1 week. --- # Qdrant Model Migration > Zero-downtime embedding switch. Named-vector add/delete needs v1.18+ **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-model-migration/SKILL.md --- ## When it is useful Zero-downtime embedding model switch in Qdrant. Named-vector add/delete needs v1.18+. v1.17 needs a new collection. Not a Pinecone namespace rewrite. ## The Catch Zero-downtime embedding switch. Named-vector add/delete needs v1.18+; v1.17 needs a new collection. Not Pinecone namespace rewrite. Distinct from pinecone-query. --- # Qdrant Multitenancy > Default payload-filter isolation, NOT collection-per-tenant **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-multitenancy/SKILL.md --- ## When it is useful Tenant isolation in Qdrant. Default is payload-filter isolation, NOT collection-per-tenant. Use when isolating customer data in one deployment. ## The Catch Default payload-filter isolation, NOT collection-per-tenant. Distinct from pinecone-cli namespaces and from model-migration. Cloud Free unused clusters suspend after 1 week. --- # Qdrant Version Upgrade > Storage compat one minor only. Cloud automates; self-host stepwise **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-version-upgrade/SKILL.md --- ## When it is useful Upgrade Qdrant server and SDKs. Storage compatibility is one minor only. Cloud automates skips; self-host must go stepwise. Zero-downtime needs replication >=2. ## The Catch Storage compat one minor only. Cloud automates skips; self-host must do stepwise. Zero-downtime needs replication >=2. Distinct from model-migration. --- # Qdrant Sizing > Cloud Free 1 GB will fail production assumptions. Not Pinecone RU/WU **Subcategory:** Qdrant **License:** Apache-2.0 **Source:** https://github.com/qdrant/skills/blob/main/skills/qdrant-sizing/SKILL.md --- ## When it is useful Size a Qdrant deployment (RAM, disk, CPU, nodes) before provisioning. Cloud Free is 1 GB RAM / 0.5 vCPU / 4 GB disk / 1 node and will fail production assumptions. Not a Pinecone RU/WU calculator. ## The Catch Cloud Free 1 GB will fail production assumptions. Not Pinecone RU/WU calculator. Unused free clusters suspended after 1 week, deleted after 4 weeks. Distinct from deployment-options. --- # Algolia Quickstart > Account/app from zero. Not index ops on existing app **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-quickstart/SKILL.md --- ## When it is useful Set up Algolia from scratch: account, CLI login, provision an application, App ID and API key. Not index or data operations on an existing authenticated app. Vendor signup steal. ## The Catch Account/app from zero. Not index ops on existing app. Vendor signup steal. Free plan: 50k records, 10k search req/mo, 5k recommend/mo, 10KB/record, 1GB/app, 1GB/index, max 1 application. Over Usage Limit may suspend OR inject Algolia attribution. Inactive >30 days may delete data. --- # Algolia Index Configuration > Guidance-only. Live writes via cli/mcp. Not Elasticsearch mapping **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-index-configuration/SKILL.md --- ## When it is useful Algolia index settings and relevance guidance (searchableAttributes, facets, ranking, synonyms). Guidance-only; live writes go through algolia-cli or MCP. Not Elasticsearch mapping. Not pinecone-full-text-search. ## The Catch Guidance-only; live writes via cli/mcp. Not Elasticsearch mapping. Distinct from pinecone-full-text-search. Free over-limit may suspend or inject attribution. --- # Algolia InstantSearch UI > JS/React/Vue. Angular InstantSearch deprecated. Not autocomplete **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-instantsearch-ui/SKILL.md --- ## When it is useful Build Algolia InstantSearch results pages in JS/React/Vue. Angular InstantSearch is deprecated. Not autocomplete/typeahead (algolia-autocomplete). Not index configuration. ## The Catch JS/React/Vue. Angular InstantSearch deprecated. Not autocomplete. Distinct from algolia-index-configuration. Free 10k search req/mo. --- # Algolia Autocomplete > Typeahead not InstantSearch SRP. False-fires on any search box **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-autocomplete/SKILL.md --- ## When it is useful Algolia Autocomplete typeahead, query suggestions, federated panels. Not InstantSearch search-results pages. False-fires on any search box. Distinct from algolia-instantsearch-ui. ## The Catch Typeahead not InstantSearch SRP. False-fires on any search box. Distinct from instantsearch-ui. Free 10k search req/mo; over-limit may suspend or inject attribution. --- # Algolia Crawler > Crawl into Algolia index, not markdown dump. Over-limit = suspend or attribution **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-crawler/SKILL.md --- ## When it is useful Crawl pages into an Algolia index via the Algolia CLI. Not a markdown dump. Not Firecrawl extract. Do not claim 5k crawls/mo (not on the Free plan policy page). ## The Catch Crawl into Algolia index, not markdown dump. Free over-limit may suspend OR inject Algolia attribution. Inactive >30 days may delete data. Do not claim 5k crawls/mo. --- # Algolia Migration > SDK major v3 to v4 / v4 to v5. Dart excluded. Not data/index migration **Subcategory:** Algolia **License:** MIT **Source:** https://github.com/algolia/skills/blob/main/skills/algolia-migration/SKILL.md --- ## When it is useful Migrate Algolia API client code across major SDK versions (v3 to v4, v4 to v5). Dart is excluded. Not data or index migration. Distinct from sanity-migration and from algolia-quickstart. ## The Catch SDK major v3 to v4 / v4 to v5. Dart excluded. Not data/index migration. Distinct from pinecone-full-text-search and from sanity-migration. --- # DuckDB Install Extension > INSTALL extension, not brew-installing DuckDB **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/install-duckdb/SKILL.md --- ## When it is useful Install or update DuckDB extensions (INSTALL / name@repo). Not first-time CLI or brew install of DuckDB itself. False-fires on first-time CLI setup. Local OSS 0 USD, no cloud, no key. ## The Catch INSTALL extension NOT brew-installing DuckDB. False-fires on first-time CLI setup. Local OSS 0 USD, no cloud, no key. Distinct from Neon/PlanetScale SQL. --- # DuckDB Read File > whats in this file steal vs Read tool. Not for source code **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/read-file/SKILL.md --- ## When it is useful Read data files (CSV, JSON, Parquet, Avro, Excel, spatial, SQLite) or remote URLs with DuckDB. Steals whats in this file vs the Read tool. Not for source code. Distinct from duckdb-convert-file. ## The Catch whats in this file steal vs Read tool. Not for source code. Local OSS 0 USD, no cloud, no key. Distinct from Neon SQL and from convert-file. --- # DuckDB Convert File > Generic convert steal (pandoc/ffmpeg). Data files only **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/convert-file/SKILL.md --- ## When it is useful Convert data files with DuckDB (CSV, Parquet, JSON, Excel, GeoJSON). Generic convert steal vs pandoc/ffmpeg. Data files only, not documents or video. ## The Catch Generic convert steal (pandoc/ffmpeg). Data files only. Distinct from duckdb-read-file. Local OSS 0 USD, no key. --- # DuckDB Attach DB > Session-state is Claude-skill local, not DuckDB server attach **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/attach-db/SKILL.md --- ## When it is useful Attach a DuckDB database file and write a SQL state file so later queries restore the session. Session-state convention is Claude-skill local, not DuckDB server ATTACH. ## The Catch Session-state convention is Claude-skill local, not DuckDB server attach. Distinct from read-file. Local OSS 0 USD, no cloud, no key. Distinct from Neon/PlanetScale. --- # DuckDB S3 Explore > Query s3:// r2:// gs:// without download. Needs provider creds. Not AWS CLI ls **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/s3-explore/SKILL.md --- ## When it is useful Explore and query s3:// r2:// gs:// without downloading. Needs provider creds. Not AWS CLI ls. DuckDB itself is local OSS 0 USD. ## The Catch Query s3:// r2:// gs:// without download. Needs provider creds. Not AWS CLI ls. Distinct from read-file local paths. DuckDB is 0 USD local; object-store bills separately. --- # DuckDB Spatial > Geo steal. Spatial extension must already be installed **Subcategory:** DuckDB **License:** MIT **Source:** https://github.com/duckdb/duckdb-skills/blob/main/skills/spatial/SKILL.md --- ## When it is useful Spatial questions in DuckDB: locations, distances, GeoJSON, Shapefile, GeoParquet. Geo steal. Spatial extension must already be installed (use duckdb-install-duckdb first). HOLD query, duckdb-docs, read-memories. ## The Catch Geo steal. Spatial extension must already be installed. Distinct from install-duckdb and from s3-explore. Local OSS 0 USD, no key. --- # Knock Setup > Vendor onboarding. Always asks for Knock account first **Subcategory:** Knock **License:** MIT **Source:** https://github.com/knocklabs/skills/blob/main/skills/knock-setup/SKILL.md --- ## When it is useful End-to-end Knock setup: connect tooling, design notification workflows, wire triggers. Always asks for a Knock account first. Not generic notifications and not Resend send-API. ## The Catch Vendor onboarding. Always asks for Knock account first. Not generic notifications. Distinct from resend-cli. Developer 0 USD/mo: 10k messages, 500 guide AU. A message = one user times one channel. Downstream SendGrid/Twilio still billed. --- # Knock In-App UI > React-only. Developer cap 500 guide AU. Branding on feeds is Starter-gated **Subcategory:** Knock **License:** MIT **Source:** https://github.com/knocklabs/skills/blob/main/skills/knock-in-app-ui/SKILL.md --- ## When it is useful Knock in-app UI in React: feeds and guides. React-only. Developer cap 500 guide active users. Knock branding on feeds is Starter-gated (250 USD/mo). Not knock-setup onboarding. ## The Catch React-only. Developer cap 500 guide AU. Branding on feeds is Starter-gated (250 USD/mo). Distinct from knock-setup and from react-email. --- # Knock CLI > Pull/push workflows via CLI. Not MCP path **Subcategory:** Knock **License:** MIT **Source:** https://github.com/knocklabs/skills/blob/main/skills/knock-cli/SKILL.md --- ## When it is useful Knock CLI to pull/push workflows, templates, and notification resources. Not the MCP path. Distinct from knock-setup and from resend-cli. ## The Catch Pull/push workflows via CLI. Not MCP path. Distinct from resend-cli. Developer 10k messages/mo; a message = one user times one channel. Downstream SendGrid/Twilio still billed. --- # Knock Migrate to Knock > Discovery-only. Writes knock-plan.md. Vendor steal vs Resend/SendGrid **Subcategory:** Knock **License:** MIT **Source:** https://github.com/knocklabs/skills/blob/main/skills/knock-migrate-to-knock/SKILL.md --- ## When it is useful Discover existing messaging (Braze, Courier, Customer.io, Iterable, SendGrid, custom) and map it to Knock. Discovery-only: will not mutate Knock resources. Writes knock-plan.md. Vendor steal vs Resend/SendGrid. HOLD lifecycle-opportunities, notification-best-practices, product-messaging-strategy. ## The Catch Discovery-only will not mutate but vendor steal vs Resend/SendGrid. Writes knock-plan.md. Distinct from knock-setup and from resend-cli. No Grafana dump. --- # chDB DataStore > Pandas API on in-process ClickHouse. Not a server. Distinct from chdb-sql **Subcategory:** ClickHouse **License:** Apache-2.0 **Source:** https://github.com/ClickHouse/agent-skills/blob/main/skills/chdb-datastore/SKILL.md --- ## When it is useful Use chDB DataStore as a pandas-compatible API backed by in-process ClickHouse. Change import pandas as pd to import chdb.datastore as pd. Not raw SQL (chdb-sql) and not a ClickHouse server. Python 3.9+, pip install chdb. ## The Catch In-process pandas twin, not a ClickHouse server. OSS/chDB is 0 USD. ClickHouse Cloud is a trial plus 300 USD credits, not a hobby plan. Do not invent a 30-day window. Distinct from chdb-sql. --- # chDB SQL > In-process ClickHouse SQL. Not DataStore pandas. No server **Subcategory:** ClickHouse **License:** Apache-2.0 **Source:** https://github.com/ClickHouse/agent-skills/blob/main/skills/chdb-sql/SKILL.md --- ## When it is useful Run ClickHouse SQL in-process via chdb.query() / Session on local files, S3, Postgres, MySQL, Iceberg, Delta with no server. Not pandas DataStore chaining (chdb-datastore) and not ClickHouse server admin (infra-clickhouse). ## The Catch Embedded SQL, not pandas DataStore. OSS/chDB 0 USD. Cloud is trial plus 300 USD credits, not hobby. Distinct from chdb-datastore and from infra-clickhouse. HOLD troubleshooting/best-practices/architecture-advisor. --- # ClickHouse JS Node Coding > @clickhouse/client Node only. Not client-web / Edge / Workers **Subcategory:** ClickHouse **License:** Apache-2.0 **Source:** https://github.com/ClickHouse/agent-skills/blob/main/skills/clickhouse-js-node-coding/SKILL.md --- ## When it is useful Write idiomatic Node.js code with @clickhouse/client (Next Node runtime, RSC, Server Actions). Never @clickhouse/client-web. Not browser, Edge, or Cloudflare Workers. Not RowBinary codecs (clickhouse-js-node-rowbinary). HOLD clickhouse-js-node-troubleshooting. ## The Catch Node-only @clickhouse/client. False-fires on Edge/Workers/client-web. Distinct from rowbinary and from chdb-sql. Cloud is trial plus 300 USD credits, not a hobby plan. --- # ClickHouse JS Node RowBinary > RowBinary codec generator. Not JSONEachRow. Node only **Subcategory:** ClickHouse **License:** Apache-2.0 **Source:** https://github.com/ClickHouse/agent-skills/blob/main/skills/clickhouse-js-node-rowbinary/SKILL.md --- ## When it is useful Generate TypeScript readers/writers for ClickHouse RowBinary / RowBinaryWithNames / RowBinaryWithNamesAndTypes on Node. Prefer JSONEachRow for string-heavy results (clickhouse-js-node-coding). Not browser. HOLD troubleshooting. ## The Catch RowBinary codecs, not JSONEachRow. Node only. Distinct from clickhouse-js-node-coding. Cloud is trial plus 300 USD credits, not hobby. Do not invent a 30-day window. --- # Infra ClickHouse > clickhousectl local vs Cloud. Cloud is trial plus credits, not hobby **Subcategory:** ClickHouse **License:** Apache-2.0 **Source:** https://github.com/ClickHouse/agent-skills/blob/main/skills/infra-clickhouse/SKILL.md --- ## When it is useful Set up ClickHouse with clickhousectl: local server for development, or managed ClickHouse Cloud for production. Default to local; ask before creating Cloud. Not chDB in-process and not JS client coding. HOLD architecture-advisor, best-practices, troubleshooting. ## The Catch clickhousectl. Local OSS 0 USD. Cloud is a trial plus 300 USD credits, NOT a hobby plan. Do not invent a 30-day window. Distinct from chdb-sql and from clickhouse-js-node-coding. --- # MotherDuck Connect > Postgres endpoint false-fires as Neon/PlanetScale. Not local DuckDB **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-connect/SKILL.md --- ## When it is useful Connect an app to MotherDuck. Recommended path is the Postgres-compatible endpoint, which false-fires as Neon or PlanetScale. Native DuckDB API is not the default. Not local DuckDB (duckdb-*). Lite 0 USD/mo: 10 GB, 10 CU hours Pulse, 3 users / 2 service accounts. ## The Catch Postgres endpoint false-fires as Neon/PlanetScale. Recommended path is not local DuckDB. Distinct from duckdb-attach-db. Lite 0 USD/mo: 10 GB storage, 10 CU hours Pulse, 3 users / 2 service accounts. Pulse only on Lite. --- # MotherDuck CLI > Not MCP. Dives/Flights are MotherDuck-specific. Lite Flights 30 min/day **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-cli/SKILL.md --- ## When it is useful Operate MotherDuck from a terminal: auth, query, Dive/Flight authoring. Not MCP (chat-only exploration). Dives and Flights are MotherDuck-specific. Lite Flights 30 min/day without CC. Distinct from duckdb-install-duckdb. ## The Catch Not MCP. Dives/Flights are MotherDuck-specific. Lite Flights 30 min/day without CC. Distinct from duckdb-* CLI. Lite 0 USD/mo still 10 GB / 10 CU hours Pulse / 2 service accounts. --- # MotherDuck Load Data > DuckDB convert/read-file twin but destination is MotherDuck. 10 GB Lite cap **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-load-data/SKILL.md --- ## When it is useful Load data into MotherDuck from files, object storage, HTTPS, dataframes, or external DBs. Twin of duckdb-convert-file / duckdb-read-file but destination is MotherDuck. Hits the Lite 10 GB cap. Not duckdb-s3-explore local. ## The Catch DuckDB convert/read-file twin but destination MotherDuck. Lite 10 GB storage cap. Distinct from duckdb-convert-file and from duckdb-s3-explore. Pulse CU hours still apply. --- # MotherDuck Share Data > Zero-copy metadata clones. Not S3 sharing. Not DuckDB ATTACH of a file **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-share-data/SKILL.md --- ## When it is useful Create MotherDuck data shares: zero-copy, read-only metadata clones of a database. Not S3 sharing and not DuckDB ATTACH of a file (duckdb-attach-db). Recipients ATTACH the share read-only. ## The Catch Zero-copy metadata clones, not S3 sharing, not DuckDB ATTACH of a file. Distinct from duckdb-attach-db and from motherduck-load-data. Lite still 3 users / 2 service accounts. --- # MotherDuck DuckLake > Default is native storage. Over-fires on lakehouse / Iceberg-on-S3 **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-ducklake/SKILL.md --- ## When it is useful Decide when DuckLake is the right MotherDuck storage vs native MotherDuck (the default). Over-fires on lakehouse / Iceberg-on-S3. Native storage is often faster for reads. Not motherduck-load-data ingestion. ## The Catch Default native storage; over-fires on lakehouse / fights Iceberg-on-S3. Distinct from duckdb-spatial and from motherduck-load-data. Lite 10 GB still applies to native storage. --- # MotherDuck REST API > Control plane, not SQL. Lite service-account cap is 2 **Subcategory:** MotherDuck **License:** MIT **Source:** https://github.com/motherduckdb/agent-skills/blob/main/skills/motherduck-rest-api/SKILL.md --- ## When it is useful MotherDuck REST control-plane admin (tokens, orgs, service accounts). Not SQL. Uses MOTHERDUCK_ADMIN_TOKEN, never for database connections (motherduck-connect). Lite service-account cap is 2. HOLD query, duckdb-sql, explore, migrate-to-motherduck, pricing-roi, partner-delivery, build-*. ## The Catch Control plane not SQL. Lite service-account cap is 2. Distinct from motherduck-connect and from duckdb-*. No 21-file dump. --- # Hugging Face Spaces > Host Gradio/Docker/Static. Free: Static + max 2 ZeroGPU. cpu-basic not free **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-spaces/SKILL.md --- ## When it is useful Host apps on Hugging Face Spaces (Gradio, Docker, or Static). Overlaps huggingface-gradio for UI — this is hosting. Free account: Static plus max 2 ZeroGPU only. cpu-basic is NOT a free fallback. Distinct from huggingface-lora-space-builder (publishes a LoRA demo). Do NOT dump hf-cloud-sagemaker. ## The Catch Hosting, not Gradio UI. Free: Static + max 2 ZeroGPU only; cpu-basic is NOT a free fallback. Distinct from huggingface-gradio. Gradio/Docker need PRO except the two ZeroGPU slots. --- # Hugging Face ZeroGPU > @spaces.GPU / pickle isolation. Gradio ZeroGPU only — Docker/Static will not schedule **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-zerogpu/SKILL.md --- ## When it is useful Write or review @spaces.GPU ZeroGPU code: pickle isolation, duration/quota, no torch.compile. Gradio ZeroGPU only — Docker, Static, and Streamlit-as-Docker will not schedule. Distinct from huggingface-spaces hosting and from huggingface-gradio UI. ## The Catch @spaces.GPU / pickle isolation. Gradio ZeroGPU only — Docker/Static/Streamlit-as-Docker will not schedule. Distinct from huggingface-spaces. Free still max 2 ZeroGPU Spaces. --- # Hugging Face LoRA Space Builder > Publishes a real private Space for a Hub LoRA. Steals make a Space/demo **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/huggingface-lora-space-builder/SKILL.md --- ## When it is useful Build and publish a Gradio Space for a user-provided Hub LoRA (private by default, ZeroGPU). Steals make a Space/demo vs huggingface-spaces. Output is a published Space, not a local script. Distinct from huggingface-zerogpu coding rules. ## The Catch Publishes a real private Space for a Hub LoRA. make a Space/demo steal vs huggingface-spaces. Distinct from huggingface-zerogpu. Free ZeroGPU cap still 2 Spaces. --- # Hugging Face Train Sentence Transformers > Router not a manual. Skip references/ and trainers break. Not generic embedding **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/train-sentence-transformers/SKILL.md --- ## When it is useful Train or fine-tune sentence-transformers (SentenceTransformer, CrossEncoder, SparseEncoder, MultiVectorEncoder). This SKILL.md is a router — skip references/ and trainers break. Not a generic embedding skill. Distinct from huggingface-trl-training. ## The Catch Router-not-manual; skip references/ and trainers break. Not generic embedding. Distinct from huggingface-trl-training and from huggingface-vision-trainer. GPU assumed, not a 0 USD laptop. --- # Hugging Face TRL Training > TRL CLI SFT/DPO/GRPO. Not huggingface-llm-trainer. GPU assumed, not 0 USD laptop **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/trl-training/SKILL.md --- ## When it is useful Train with TRL CLI (trl sft / dpo / grpo / kto / rloo / reward). Not huggingface-llm-trainer. GPU assumed, not a 0 USD laptop. Distinct from huggingface-train-sentence-transformers and from huggingface-vision-trainer. ## The Catch TRL CLI SFT/DPO/GRPO. Not huggingface-llm-trainer. GPU assumed; not a 0 USD laptop. Distinct from train-sentence-transformers. Cloud GPU costs apply. --- # Hugging Face hf-mem > VRAM estimate via HTTP Range, no download. Needs uvx. Gated models need HF_TOKEN **Subcategory:** Hugging Face **License:** Apache-2.0 **Source:** https://github.com/huggingface/skills/blob/main/skills/hf-mem/SKILL.md --- ## When it is useful Estimate VRAM/memory to load Safetensors or GGUF from the Hub via HTTP Range — no weight download. Needs uvx. Gated/private models need HF_TOKEN. Distinct from huggingface-local-models (actually running GGUF). ## The Catch VRAM estimate via HTTP Range, no download. Needs uvx. Gated models need HF_TOKEN. Distinct from huggingface-local-models. Does not train or host. --- # Novu Trigger Notification > knock-setup twin. Topic fan-out bills one run per matching subscriber **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/trigger-notification/SKILL.md --- ## When it is useful Trigger Novu workflows (single, bulk, broadcast, topic). knock-setup twin for sending. Topic fan-out bills one workflow run per matching subscriber vs Free 10,000 runs/mo. Meter is workflow runs, not Knock messages. Downstream email/SMS still billed. Distinct from knock-setup. ## The Catch knock-setup twin. Topic fan-out bills one run per matching subscriber vs 10k Free runs/mo. Meter is workflow runs, NOT Knock messages. Over-limit does not throttle. Distinct from knock-setup. Downstream SendGrid/Twilio still billed. --- # Novu Manage Subscribers > Not user-auth. subscriberId is not a Clerk/Auth.js user **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/manage-subscribers/SKILL.md --- ## When it is useful Create, update, search, delete Novu subscribers and topics. Not user-auth — subscriberId is not a Clerk or Auth.js user. Distinct from novu-trigger-notification and from knock-setup identify. ## The Catch Not user-auth; subscriberId ≠ Clerk/Auth.js user. Distinct from knock-setup and from novu-trigger-notification. Cloud Free 10k workflow runs/mo still applies when you later trigger. --- # Novu Inbox Integration > knock-in-app-ui twin. Feeds, not Knock guides. HMAC required in production **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/inbox-integration/SKILL.md --- ## When it is useful Integrate Novu in-app Inbox (bell, feed, preferences) in React/Next/vanilla JS. knock-in-app-ui twin but feeds, not Knock guides. HMAC required in production. Distinct from knock-in-app-ui. ## The Catch knock-in-app-ui twin (feeds, not Knock guides). Distinct from knock-in-app-ui. HMAC required in production. Cloud Free 10k workflow runs/mo; in-app still consumes a run. OSS self-host 0 USD. --- # Novu Manage Preferences > Not Knock preferences API. Two-level model only (workflow + subscriber) **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/manage-preferences/SKILL.md --- ## When it is useful Configure Novu notification preferences: workflow defaults plus subscriber overrides. Two-level model only. Not the Knock preferences API. Distinct from novu-inbox-integration UI and from knock-in-app-ui. ## The Catch Not Knock preferences API; two-level model only. Distinct from knock-in-app-ui and from novu-inbox-integration. readOnly hides UI; critical is the runtime bypass. Cloud Free 10k runs/mo. --- # Novu Framework Integration > @novu/framework + Bridge. Cloud still executes Bridge. Not Resend-only react-email **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/framework-integration/SKILL.md --- ## When it is useful Build code-first Novu workflows with @novu/framework and a Bridge Endpoint. Cloud still executes the Bridge. Can render React Email but is not Resend-only react-email. Distinct from novu-dashboard-workflows and from react-email. HOLD novu-design-workflow. Skip connect-agent. ## The Catch @novu/framework + Bridge; Cloud still executes Bridge. Not Resend-only react-email. Distinct from novu-dashboard-workflows and from react-email. Free 10k workflow runs/mo; over-limit does not throttle. --- # Novu Dashboard Workflows > MCP/dashboard step content, not framework-integration **Subcategory:** Novu **License:** MIT **Source:** https://github.com/novuhq/skills/blob/main/skills/dashboard-workflows/SKILL.md --- ## When it is useful Author step content (subject, body, editorType, conditions) for Novu Dashboard or MCP workflows. Not @novu/framework (novu-framework-integration). HOLD novu-design-workflow. Skip connect-agent. ## The Catch MCP/dashboard step content, not framework-integration. Distinct from novu-framework-integration and from knock-cli. Cloud Free 10k workflow runs/mo, no CC. OSS self-host 0 USD. --- # Trigger Getting Started > Vendor lock. CLI login + TRIGGER_SECRET_KEY. Self-contained **Subcategory:** Trigger.dev **License:** Apache-2.0 **Source:** https://github.com/triggerdotdev/trigger.dev/blob/main/packages/cli-v3/skills/trigger-getting-started/SKILL.md --- ## When it is useful Bootstrap Trigger.dev into an existing project: CLI login, install @trigger.dev/sdk, write trigger.config.ts, scaffold /trigger, set TRIGGER_SECRET_KEY. Self-contained (runs before SDK install). Not Inngest setup. Distinct from trigger-authoring-tasks. ## The Catch Vendor lock. Needs CLI login + TRIGGER_SECRET_KEY from the dashboard (human steps). Free 0 USD/mo: 5 USD credits/mo, 20 concurrent runs, 10 Realtime connections, 10 schedules, 1-day logs. Do not invent hard-stop phrasing beyond Free needing an upgrade after credits. Distinct from inngest-setup. --- # Trigger Authoring Tasks > CLI stub; 404s until @trigger.dev/sdk installed. False-fires on BullMQ/Inngest **Subcategory:** Trigger.dev **License:** Apache-2.0 **Source:** https://github.com/triggerdotdev/trigger.dev/blob/main/packages/cli-v3/skills/trigger-authoring-tasks/SKILL.md --- ## When it is useful Write Trigger.dev tasks (task/schemaTask, retries, waits, queues, cron). CLI stub — full guide lives in node_modules/@trigger.dev/sdk and 404s until the SDK is installed. False-fires on BullMQ/Inngest. Never import the task instance into a route handler (type-only + tasks.trigger). Distinct from inngest-durable-functions. ## The Catch CLI stub; 404s until @trigger.dev/sdk is installed. False-fires on BullMQ/Inngest. Never import the task instance into a route handler. Distinct from inngest-durable-functions. Free 20 concurrent runs / 5 USD credits/mo. --- # Trigger Authoring Chat Agent > CLI stub. Loud vs Vercel useChat / tanstack-ai-chat-experience **Subcategory:** Trigger.dev **License:** Apache-2.0 **Source:** https://github.com/triggerdotdev/trigger.dev/blob/main/packages/cli-v3/skills/trigger-authoring-chat-agent/SKILL.md --- ## When it is useful Build durable AI chat agents with Trigger.dev chat.agent / toStreamTextOptions. CLI stub until SDK installed. Loud vs Vercel useChat and tanstack-ai-chat-experience. Forgetting ...chat.toStreamTextOptions() silently no-ops compaction/steering. Distinct from inngest-agents. ## The Catch CLI stub. Loud vs Vercel useChat / tanstack-ai-chat-experience. Forgetting ...chat.toStreamTextOptions() silently no-ops compaction/steering. Distinct from inngest-agents. Free still 5 USD credits/mo. --- # Trigger Chat Agent Advanced > CLI stub. Version-gate. Re-POSTing /sessions drops the message **Subcategory:** Trigger.dev **License:** Apache-2.0 **Source:** https://github.com/triggerdotdev/trigger.dev/blob/main/packages/cli-v3/skills/trigger-chat-agent-advanced/SKILL.md --- ## When it is useful Advanced Trigger.dev chat.agent patterns: sessions, human-in-the-loop, sub-agents, compaction, fast starts. CLI stub until SDK installed. Version-gate. Re-POSTing /sessions drops the message. Distinct from trigger-authoring-chat-agent basics. HOLD trigger-cost-savings. ## The Catch CLI stub. Version-gate. Re-POSTing /sessions drops the message. Distinct from trigger-authoring-chat-agent and from inngest-agents. Free 5 USD credits/mo; do not invent hard-stop phrasing. --- # Trigger Realtime and Frontend > CLI stub. Loud vs inngest-realtime. Free cap 10 concurrent Realtime **Subcategory:** Trigger.dev **License:** Apache-2.0 **Source:** https://github.com/triggerdotdev/trigger.dev/blob/main/packages/cli-v3/skills/trigger-realtime-and-frontend/SKILL.md --- ## When it is useful Subscribe to Trigger.dev runs from React/Next (useRealtimeRun, useTaskTrigger) and mint scoped public tokens. CLI stub until SDK installed. Loud vs inngest-realtime. createPublicToken cannot trigger — use createTriggerPublicToken. Free cap 10 concurrent Realtime connections. HOLD trigger-cost-savings. ## The Catch CLI stub. Loud vs inngest-realtime. Free cap 10 concurrent Realtime connections. createPublicToken cannot trigger. Distinct from inngest-realtime. Free also 20 concurrent runs / 5 USD credits/mo / 1-day logs. --- # Axiom Alerting > v2 monitors Bearer from .axiom.toml not Knock **Subcategory:** Axiom **License:** MIT **Source:** https://github.com/axiomhq/skills/blob/main/skills/axiom-alerting/SKILL.md --- ## When it is useful Build Axiom v2 API monitors and alerts. Distinct from Knock Novu. ## The Catch Personal 0 USD permanent 500GB load 10 GB-hours query 25GB storage 30-day. Bearer from .axiom.toml. --- # Axiom Building Dashboards > False-fires on Grafana Datadog Splunk migrate **Subcategory:** Axiom **License:** MIT **Source:** https://github.com/axiomhq/skills/blob/main/skills/axiom-building-dashboards/SKILL.md --- ## When it is useful Build Axiom dashboards. False-fires on Grafana Datadog. Splunk migrate overlap with spl-to-apl. ## The Catch Personal 0 USD permanent no CC. Distinct from axiom-alerting and spl-to-apl overlap note. --- # Axiom Query Metrics > otel metrics v1 scripts 404 if only SKILL.md **Subcategory:** Axiom **License:** MIT **Source:** https://github.com/axiomhq/skills/blob/main/skills/axiom-query-metrics/SKILL.md --- ## When it is useful Query otel:metrics:v1 datasets with MPL. Scripts relative to skill folder 404 if only SKILL.md copied. ## The Catch Personal 0 USD permanent. Dataset kind otel:metrics:v1. Scripts need full skill folder not SKILL.md alone. --- # Axiom Metrics Chart > Consumer of query-metrics zero-dep Unicode **Subcategory:** Axiom **License:** MIT **Source:** https://github.com/axiomhq/skills/blob/main/skills/axiom-metrics-chart/SKILL.md --- ## When it is useful Render Axiom metrics charts. Consumer of query-metrics. Zero-dep Unicode; PNG SVG needs gnuplot. ## The Catch Personal 0 USD. Zero-dep Unicode charts. PNG SVG needs gnuplot. Distinct from query-metrics. --- # Axiom SPL to APL > Join preview 50k time pickers dont translate **Subcategory:** Axiom **License:** MIT **Source:** https://github.com/axiomhq/skills/blob/main/skills/axiom-spl-to-apl/SKILL.md --- ## When it is useful Translate Splunk SPL to Axiom APL. Join preview 50k rows. Time pickers do not translate. Distinct from ClickHouse MotherDuck SQL. ## The Catch Personal 0 USD permanent 30-day retention. Join preview 50k. Distinct from chdb-sql and duckdb SQL. --- # Contentful API > Five APIs curl-only false-fires vs JS SDK **Subcategory:** Contentful **License:** MIT **Source:** https://github.com/contentful/skills/blob/main/skills/contentful-api/SKILL.md --- ## When it is useful Call Contentful CMA CDA CPA Images GraphQL via curl. False-fires when they wanted JS SDK nextjs. ## The Catch MIT. Pricing not verified this pass do not invent free-tier numbers. Distinct from contentful-nextjs and sanity. --- # Contentful Next.js > Next-only steals Remix Astro Nuxt vs Sanity **Subcategory:** Contentful **License:** MIT **Source:** https://github.com/contentful/skills/blob/main/skills/contentful-nextjs/SKILL.md --- ## When it is useful Integrate Contentful into Next.js only. Steals Remix Astro Nuxt. Distinct from Sanity portable-text Next. ## The Catch MIT. Pricing not verified this pass. Distinct from contentful-api and sanity-next. --- # Contentful Migration > False-fires on Prisma Drizzle vs sanity-migration **Subcategory:** Contentful **License:** MIT **Source:** https://github.com/contentful/skills/blob/main/skills/contentful-migration/SKILL.md --- ## When it is useful Write Contentful content-model migrations. False-fires on Prisma Drizzle. Distinct from sanity-migration. ## The Catch MIT. Pricing not verified this pass. Distinct from sanity-migration and contentful-api. --- # Contentful Custom App From Scratch > App Framework only not marketing site **Subcategory:** Contentful **License:** MIT **Source:** https://github.com/contentful/skills/blob/main/skills/contentful-custom-app-from-scratch/SKILL.md --- ## When it is useful Scaffold Contentful App Framework custom apps with create-contentful-app. App Framework only not marketing site. ## The Catch MIT. Pricing not verified this pass. App Framework only. HOLD enhancement guide personalization. --- # Deno Deploy > Deploy Free 1M req 20GiB egress not 100GB **Subcategory:** Deno **License:** MIT **Source:** https://github.com/denoland/skills/blob/main/skills/deno-deploy/SKILL.md --- ## When it is useful Deploy apps to Deno Deploy Free tier. Parent deno HOLD. Sandboxes are Pro not Free. ## The Catch Deploy Free 1M req 20 GiB egress 10 CPU-hr 150 GiB-hr mem 10 apps 5 domains KV 1 GiB. No pause-when-over claim. --- # Deno Frontend > Next-under-Deno false-fires vs Node Next **Subcategory:** Deno **License:** MIT **Source:** https://github.com/denoland/skills/blob/main/skills/deno-frontend/SKILL.md --- ## When it is useful Build frontend with Deno React Vite Astro SvelteKit Next or Fresh 2.x. Parent deno HOLD. ## The Catch Next-under-Deno false-fires vs Node Next. Fresh 2.x. Deploy Free 1M/20GiB/10 CPU-hr. --- # Migrate to Deno > Node runtime port false-fires **Subcategory:** Deno **License:** MIT **Source:** https://github.com/denoland/skills/blob/main/skills/migrate-to-deno/SKILL.md --- ## When it is useful Port Node apps onto Deno runtime. Parent HOLD. ## The Catch Distinct from deno-frontend and deno-deploy. --- # Deno Sandbox > LOUD Sandboxes not on Free Pro 20 USD **Subcategory:** Deno **License:** MIT **Source:** https://github.com/denoland/skills/blob/main/skills/deno-sandbox/SKILL.md --- ## When it is useful Run untrusted AI code in Deno Sandboxes @deno/sandbox. Cloudflare sandbox-sdk twin. Parent deno HOLD. ## The Catch LOUD: Sandboxes NOT on Free. Pro 20 USD/mo. Cloudflare sandbox-sdk twin. Distinct from deno-deploy. --- # Logfire Instrumentation > configure vs instrument ordering silently drops traces **Subcategory:** Logfire **License:** MIT **Source:** https://github.com/pydantic/logfire/blob/main/docs/skills/logfire-instrumentation/SKILL.md --- ## When it is useful Instrument apps with Logfire. Vendor lock. configure() vs instrument_*() ordering silently drops traces. False-fires on OTel Sentry Axiom. ## The Catch Personal Free forever 10M records/mo hard-capped 0 USD 3 projects 1 seat 2 guests 30-day. Distinct from axiom. --- # Logfire Query > Personal Query API le 500 per day **Subcategory:** Logfire **License:** MIT **Source:** https://github.com/pydantic/skills/blob/main/skills/logfire-query/SKILL.md --- ## When it is useful Query Logfire telemetry. Personal Query API le 500/day. False-fires on find root cause without Logfire. ## The Catch Personal Free forever. Query API le 500/day. Distinct from instrumentation and axiom. --- # Logfire UI > Codex-browser-shaped never-fires without browser **Subcategory:** Logfire **License:** MIT **Source:** https://github.com/pydantic/skills/blob/main/skills/logfire-ui/SKILL.md --- ## When it is useful Drive the Logfire UI. Codex-browser-shaped. Never-fires if agent has no browser. ## The Catch Personal Free forever. UI needs browser. Distinct from query and instrumentation. --- # Logfire Infrastructure > OTel Collector no app-code false-fires on k8s **Subcategory:** Logfire **License:** MIT **Source:** https://github.com/pydantic/skills/blob/main/skills/logfire-infrastructure/SKILL.md --- ## When it is useful Ship infra telemetry to Logfire via OTel Collector without app-code. False-fires on generic k8s postgres monitor. ## The Catch Personal Free forever 10M records hard-capped 0 USD. Distinct from instrumentation. No plugins twins. --- # Logfire Evals > Python-only pydantic_evals omit configure local **Subcategory:** Logfire **License:** MIT **Source:** https://github.com/pydantic/skills/blob/main/skills/logfire-evals/SKILL.md --- ## When it is useful Run Python-only pydantic_evals with Logfire. Distinct from Axiom HOLD writing-evals. Omitting logfire.configure() keeps results local. ## The Catch Personal Free forever. Python-only pydantic_evals. Distinct from axiom writing-evals HOLD. No plugins twins. --- # Mixpanel Tracking Implementation > Vendor lock distinct from install false-fires analytics **Subcategory:** Mixpanel **License:** Apache-2.0 **Source:** https://github.com/mixpanel/ai-plugins/blob/main/plugins/mixpanel/skills/tracking-implementation/SKILL.md --- ## When it is useful Implement Mixpanel tracking end-to-end. Vendor lock. Distinct from install. False-fires on generic add analytics. ## The Catch Free no trial 1M events. Over 1M Free blocks viewing reports not ingest. Distinct from PostHog. --- # Mixpanel Create Dashboard > Generic slug vs axiom dashboards Engine required **Subcategory:** Mixpanel **License:** Apache-2.0 **Source:** https://github.com/mixpanel/ai-plugins/blob/main/plugins/mixpanel/skills/create-dashboard/SKILL.md --- ## When it is useful Build Mixpanel dashboards with Engine required. Free cap 5 saved reports per user. Distinct from axiom-building-dashboards. ## The Catch Free 1M events 5 saved reports/user. Cohorts creatable not saved. Over 1M blocks viewing not ingest. --- # Mixpanel Manage Feature Flags > Not PostHog LaunchDarkly Engine required **Subcategory:** Mixpanel **License:** Apache-2.0 **Source:** https://github.com/mixpanel/ai-plugins/blob/main/plugins/mixpanel/skills/manage-feature-flags/SKILL.md --- ## When it is useful Create roll out debug Mixpanel feature flags. Not PostHog LaunchDarkly. Engine required. Do not invent Flags usage amounts. ## The Catch Free no trial as long as you want. Engine required. Distinct from PostHog flags. Do not invent Flags quotas. --- # n8n Workflow Lifecycle > Official lifecycle false-fires on ad-hoc node dumps **Subcategory:** n8n **License:** Apache-2.0 **Source:** https://github.com/n8n-io/skills/blob/main/skills/n8n-workflow-lifecycle/SKILL.md --- ## When it is useful Create activate deactivate and version n8n workflows the official way. HOLD other 8 skills. ## The Catch Official lifecycle. Cloud no 0 USD hobby Starter 20 EUR annual. CE self-host is 0 forever. --- # n8n Subworkflows > Official subworkflow false-fires on copy-paste nodes **Subcategory:** n8n **License:** Apache-2.0 **Source:** https://github.com/n8n-io/skills/blob/main/skills/n8n-subworkflows/SKILL.md --- ## When it is useful Extract and call reusable n8n subworkflows officially. HOLD other skills. ## The Catch Official subworkflows. Meter is executions. CE self-host 0 forever; Cloud Starter paid. --- # n8n Expressions > Expression false-fires on raw JS Code node **Subcategory:** n8n **License:** Apache-2.0 **Source:** https://github.com/n8n-io/skills/blob/main/skills/n8n-expressions/SKILL.md --- ## When it is useful Write n8n expressions the official way. Distinct from code-nodes. HOLD other skills. ## The Catch Official expressions vs Code node. Cloud no 0 USD hobby. CE self-host 0 forever. --- # n8n Loops > Loop false-fires on SplitInBatches misuse **Subcategory:** n8n **License:** Apache-2.0 **Source:** https://github.com/n8n-io/skills/blob/main/skills/n8n-loops/SKILL.md --- ## When it is useful Build official n8n loops and batching. HOLD other skills. ## The Catch Official loops. Distinct from expressions and code-nodes. CE self-host 0 forever. --- # n8n Code Nodes > JS/Python Code node false-fires on expressions **Subcategory:** n8n **License:** Apache-2.0 **Source:** https://github.com/n8n-io/skills/blob/main/skills/n8n-code-nodes/SKILL.md --- ## When it is useful Use n8n Code nodes in JS or Python officially. Distinct from expressions. HOLD other skills. ## The Catch Official Code nodes. Trial 1000 exec 5 concurrent 180s. CE self-host 0 forever. --- # Twilio Send Message > Vendor lock trial PAYG steal vs Knock Novu **Subcategory:** Twilio **License:** MIT **Source:** https://github.com/twilio/ai/blob/main/skills/twilio-send-message/SKILL.md --- ## When it is useful Send SMS MMS WhatsApp via Twilio Programmable Messaging. WhatsApp 24-hour window. Distinct from Knock Novu Resend. HOLD sms/whatsapp split skills. ## The Catch No 0 USD hobby. Official trial 30 days no CC then PAYG. Geo-limited templates not custom bodies. Distinct from Knock Novu. --- # Twilio Verify Send OTP > Verify product not Programmable Messaging **Subcategory:** Twilio **License:** MIT **Source:** https://github.com/twilio/ai/blob/main/skills/twilio-verify-send-otp/SKILL.md --- ## When it is useful Send OTPs with Twilio Verify not Programmable Messaging. Distinct from Infisical. Trial verified-number cap makes OTP loops fail closed. ## The Catch Verify product. Trial 30 days then PAYG. Verified-number cap fails closed. Do not invent max 5 recipients. --- # Twilio Voice Outbound Calls > Irreversible charged TCPA quiet hours **Subcategory:** Twilio **License:** MIT **Source:** https://github.com/twilio/ai/blob/main/skills/twilio-voice-outbound-calls/SKILL.md --- ## When it is useful Place outbound Twilio Voice calls. Irreversible and charged. TCPA quiet hours. Distinct from TwiML skill. ## The Catch Trial 30 days then PAYG. Charged irreversible. TCPA quiet hours. Distinct from twilio-voice-twiml. --- # Twilio Voice TwiML > Webhook TwiML not REST dial pair outbound **Subcategory:** Twilio **License:** MIT **Source:** https://github.com/twilio/ai/blob/main/skills/twilio-voice-twiml/SKILL.md --- ## When it is useful Serve webhook TwiML for Voice not REST dial. Pair with twilio-voice-outbound-calls. ## The Catch Trial 30 days then PAYG. Webhook TwiML not REST dial. Distinct from outbound-calls and Studio. --- # Twilio Studio Flows > Visual Studio vs TwiML code distinct from n8n **Subcategory:** Twilio **License:** MIT **Source:** https://github.com/twilio/ai/blob/main/skills/twilio-studio-flows/SKILL.md --- ## When it is useful Build Twilio Studio visual flows. Visual Studio vs TwiML or code. Distinct from n8n workflows. ## The Catch Trial 30 days then PAYG. Visual Studio vs TwiML. Distinct from n8n and voice-twiml. No ~45 dump. --- # Saleor Core > Internals encyclopedia version-gate 3.23 not getting-started **Subcategory:** Saleor **License:** BSD-3-Clause **Source:** https://github.com/saleor/agent-skills/blob/main/skills/saleor-core/SKILL.md --- ## When it is useful Saleor backend internals encyclopedia. Version-gate 3.23+. Not getting-started. Distinct from Shopify Contentful Sanity. ## The Catch Repo LICENSE is BSD-3-Clause skill YAML says MIT use LICENSE file. OSS free personal plus commercial. Do not invent Cloud quote pricing. --- # Saleor App > Next.js examples claim framework-agnostic vs Contentful App **Subcategory:** Saleor **License:** BSD-3-Clause **Source:** https://github.com/saleor/agent-skills/blob/main/skills/saleor-app/SKILL.md --- ## When it is useful Build Saleor apps protocol manifest appbridge webhooks. Next.js examples but claims framework-agnostic. Distinct from Contentful App Framework. ## The Catch Repo LICENSE BSD-3-Clause not skill YAML MIT. OSS free. Distinct from contentful-custom-app-from-scratch. --- # Saleor Configurator > Store config as code vs Contentful migration scripts **Subcategory:** Saleor **License:** BSD-3-Clause **Source:** https://github.com/saleor/agent-skills/blob/main/skills/saleor-configurator/SKILL.md --- ## When it is useful Manage Saleor store configuration as code. Distinct from Contentful migration scripts. HOLD storefront-builder. ## The Catch Repo LICENSE BSD-3-Clause. OSS free personal plus commercial. Distinct from contentful-migration. Do not invent Cloud pricing. --- # Saleor Storefront > GraphQL products checkout channels not Next starter **Subcategory:** Saleor **License:** BSD-3-Clause **Source:** https://github.com/saleor/agent-skills/blob/main/skills/saleor-storefront/SKILL.md --- ## When it is useful Saleor GraphQL storefront API products checkout channels. Framework-agnostic API not Next starter. HOLD storefront-builder. ## The Catch Repo LICENSE BSD-3-Clause. OSS free. Framework-agnostic GraphQL not Next starter. Distinct from Shopify. --- # Loops API > Five product surfaces in one vendor lock **Subcategory:** Loops **License:** MIT **Source:** https://github.com/loops-so/skills/blob/main/skills/loops-api/SKILL.md --- ## When it is useful Use Loops API across five product surfaces in one skill. Vendor lock. Distinct from resend-cli react-email and n8n-loops-official. ## The Catch Free up to 4000 sends/month Powered by Loops footer all features no CC. 1000 stored subscribed contacts is slider default. Distinct from n8n-loops-official. --- # Loops CLI > Overlaps loops-api distinct from resend-cli **Subcategory:** Loops **License:** MIT **Source:** https://github.com/loops-so/skills/blob/main/skills/loops-cli/SKILL.md --- ## When it is useful Use Loops CLI from the terminal. Overlaps loops-api. Distinct from resend-cli. HOLD email-sending-best-practices. ## The Catch Free 4000 sends/month no CC. Distinct from resend-cli and loops-api. HOLD loops-email-sending-best-practices. --- # Loops LMX > False-fires on generic welcome email React Email **Subcategory:** Loops **License:** MIT **Source:** https://github.com/loops-so/skills/blob/main/skills/loops-lmx/SKILL.md --- ## When it is useful Write and review Loops Markup Language email bodies. False-fires on generic welcome email or React Email. ## The Catch Free 4000 sends/month Powered by Loops footer. Distinct from react-email. False-fires on generic welcome email. --- # Temporal Developer > Vendor lock + always-on-adjacent. start-dev is not production **Subcategory:** Temporal **License:** MIT **Source:** https://github.com/temporalio/skill-temporal-developer/blob/main/SKILL.md --- ## When it is useful Build Temporal workflows, activities, and workers (Python/TS/Go/Java/.NET/Ruby/Rust), CLI, non-determinism, signals/queries. Distinct from inngest-durable-functions and trigger-authoring-tasks. ## The Catch Vendor lock + always-on-adjacent on any durable workflow. Local `temporal server start-dev` is not production. OSS self-host $0. Distinct from inngest-durable-functions / trigger-authoring-tasks. --- # Temporal Ops > CLI ops not SDK. Names temporal-workertuning. False-fires generic TLS/gRPC **Subcategory:** Temporal **License:** MIT **Source:** https://github.com/temporalio/skill-temporal-ops/blob/main/SKILL.md --- ## When it is useful Administer Temporal Cloud or self-hosted via CLI (namespaces, certs, batch cancel/reset, billing, SAML/SCIM) — not SDK code. ## The Catch Names temporal-workertuning sibling (do not invent/dump). False-fires on generic TLS/gRPC. Cloud ops paths need paid Cloud. OSS self-host $0. Cloud not hobby. --- # Temporal Cloud > Cloud-only. Not $0 — $1,000 credits expire 90 days then Essentials **Subcategory:** Temporal **License:** MIT **Source:** https://github.com/temporalio/skills-temporal-cloud/blob/main/SKILL.md --- ## When it is useful Temporal Cloud connect/auth/x509/namespace/PrivateLink/no pollers/RESOURCE_EXHAUSTED. ## The Catch Cloud-only. Loud: not $0 — $1,000 credits expire 90 days, then Essentials greater of $100/mo or 5% of usage. CC required for Cloud; OSS quickstart no CC. Keep with Catch (OSS path is developer/ops). Do not HOLD as paid-only. Distinct from Inngest/Trigger/Loops/n8n-loops-official. --- # Appwrite CLI > CLI/config/CI not SDK. Cloud Free pauses after 1 week Console inactivity **Subcategory:** Appwrite **License:** BSD-3-Clause **Source:** https://github.com/appwrite/skills/blob/main/skills/appwrite-cli/SKILL.md --- ## When it is useful Use Appwrite CLI for config, pull/push, CI, generate/types. Not SDK code. HOLD swift, go/dotnet/php/ruby/rust, deploy-function, deploy-site. ## The Catch Cloud Free $0 pauses after 1 week Console inactivity; 2 projects, 2 functions/project. OSS self-host $0 production path. Not SDK. Distinct from supabase/convex/neon. HOLD swift go/dotnet/php/ruby/rust deploy-function deploy-site. --- # Appwrite TypeScript > Web/RN/Node-Deno. TablesDB not deprecated Databases **Subcategory:** Appwrite **License:** BSD-3-Clause **Source:** https://github.com/appwrite/skills/blob/main/skills/appwrite-typescript/SKILL.md --- ## When it is useful Build Appwrite apps in web, React Native, or Node/Deno. TablesDB not deprecated Databases. Distinct from supabase JS. ## The Catch TablesDB not deprecated Databases. Distinct from supabase JS. Cloud Free $0 pauses after 1 week Console inactivity; 2 projects, 2 functions/project. OSS self-host $0. --- # Appwrite Dart > Flutter or server Dart. Distinct from supabase-flutter **Subcategory:** Appwrite **License:** BSD-3-Clause **Source:** https://github.com/appwrite/skills/blob/main/skills/appwrite-dart/SKILL.md --- ## When it is useful Build Flutter or server Dart with Appwrite. Distinct from supabase-flutter. ## The Catch Distinct from supabase-flutter. Cloud Free $0 pauses after 1 week Console inactivity; 2 projects, 2 functions/project. OSS self-host $0. --- # Appwrite Python > Django/Flask/FastAPI server SDK. Server-only no browser **Subcategory:** Appwrite **License:** BSD-3-Clause **Source:** https://github.com/appwrite/skills/blob/main/skills/appwrite-python/SKILL.md --- ## When it is useful Use Appwrite Python server SDK in Django, Flask, or FastAPI. Server-only, no browser. Distinct from supabase-py. ## The Catch Server-only no browser. Distinct from supabase-py. Cloud Free $0 pauses after 1 week Console inactivity; 2 projects, 2 functions/project. OSS self-host $0. --- # Appwrite Kotlin > Android or Kotlin/JVM. Activity OAuth/coroutines. Not Swift/KMP **Subcategory:** Appwrite **License:** BSD-3-Clause **Source:** https://github.com/appwrite/skills/blob/main/skills/appwrite-kotlin/SKILL.md --- ## When it is useful Build native Android or Kotlin/JVM with Appwrite. Activity OAuth and coroutines. Not Swift/KMP. ## The Catch Not Swift/KMP. Activity OAuth/coroutines. Cloud Free $0 pauses after 1 week Console inactivity; 2 projects, 2 functions/project. Distinct from supabase/convex/neon. OSS self-host $0. --- # Stream Builder > Only when named. Do not auto-select. Web React defaults to stream-react **Subcategory:** GetStream **License:** Apache-2.0 **Source:** https://github.com/GetStream/agent-skills/blob/main/skills/stream-builder/SKILL.md --- ## When it is useful Use only when the user names stream-builder or /stream-builder. Do NOT auto-select. Web React defaults to stream-react. ## The Catch Do not auto-select. Web React defaults to stream-react. Chat Free 1,000 MAU, 100 concurrent, no CC. Do not invent Video/Feeds caps. HOLD stream parent, stream-docs, android/swift/unity/unreal. --- # Stream Feeds Migration > Feeds v2 to v3. Version-gate. Needs live v2 key+secret. Stale when v2 is gone **Subcategory:** GetStream **License:** Apache-2.0 **Source:** https://github.com/GetStream/agent-skills/blob/main/skills/stream-feeds-migration/SKILL.md --- ## When it is useful Generate Activity Feeds v2 to v3 sync mapping. Version-gate. Needs live v2 key+secret. Stale the day v2 is gone. ## The Catch Version-gate. Needs live v2 key+secret. Stale the day v2 is gone. Do not invent Video/Feeds caps. Distinct from stream-react Feeds path. --- # Stream React > Steals generic chat app. Web only not RN. Distinct from livekit-agents Knock Novu **Subcategory:** GetStream **License:** Apache-2.0 **Source:** https://github.com/GetStream/agent-skills/blob/main/skills/stream-react/SKILL.md --- ## When it is useful Build React/Next Chat+Video+Feeds+Moderation. Steals generic chat app. Web only, not RN. Distinct from livekit-agents / Knock/Novu. ## The Catch Steals generic chat app. Web only not RN. Distinct from livekit-agents / knock-in-app-ui / novu-inbox-integration. Chat Free 1,000 MAU, 100 concurrent, no CC. Do not invent Video/Feeds caps. --- # Stream React Native > RN CLI/Expo. New Architecture required. Distinct from stream-react **Subcategory:** GetStream **License:** Apache-2.0 **Source:** https://github.com/GetStream/agent-skills/blob/main/skills/stream-react-native/SKILL.md --- ## When it is useful Build Stream Chat/Video/Feeds in React Native CLI or Expo. New Architecture required. Distinct from stream-react. ## The Catch New Architecture required. Distinct from stream-react. Chat Free 1,000 MAU, 100 concurrent, no CC. Do not invent Video/Feeds caps. --- # Stream Flutter > Flutter Chat/Video/Feeds. Feeds has no pre-built UI. Distinct from appwrite-dart **Subcategory:** GetStream **License:** Apache-2.0 **Source:** https://github.com/GetStream/agent-skills/blob/main/skills/stream-flutter/SKILL.md --- ## When it is useful Build Stream Chat/Video/Feeds in Flutter. Feeds has no pre-built UI. Distinct from appwrite-dart. ## The Catch Feeds has no pre-built UI. Distinct from appwrite-dart. Chat Free 1,000 MAU, 100 concurrent, no CC. Do not invent Video/Feeds caps. --- # MLflow Onboarding > GenAI vs classic ML get-started router **Subcategory:** MLflow **License:** Apache-2.0 **Source:** https://github.com/mlflow/skills/blob/main/mlflow-onboarding/SKILL.md --- ## When it is useful Get started or add MLflow — GenAI vs classic ML tutorial paths. ## The Catch Router-adjacent (picks tutorials). False-fires on generic add observability. Distinct from logfire-instrumentation. OSS self-host $0. Do not reprint Databricks managed DBU. --- # MLflow Instrumenting Tracing > Add MLflow Tracing across Python/TS agent stacks **Subcategory:** MLflow **License:** Apache-2.0 **Source:** https://github.com/mlflow/skills/blob/main/instrumenting-with-mlflow-tracing/SKILL.md --- ## When it is useful Add MLflow Tracing (Python/TS; LangGraph/LangChain/OpenAI/DSPy/CrewAI). Must-load before agent-evaluation (that sibling is HOLD). ## The Catch Vendor lock vs live logfire-instrumentation. HOLD agent-evaluation sibling. OSS self-host $0. --- # MLflow Retrieving Traces > Search/filter traces by ID, status, tags, latency **Subcategory:** MLflow **License:** Apache-2.0 **Source:** https://github.com/mlflow/skills/blob/main/retrieving-mlflow-traces/SKILL.md --- ## When it is useful Get/search/filter MLflow traces by ID/status/tags/latency. ## The Catch Fetch/search, not analysis — pair with mlflow-analyzing-trace. --- # MLflow Analyzing Trace > One trace ID for debug, RCA, quality **Subcategory:** MLflow **License:** Apache-2.0 **Source:** https://github.com/mlflow/skills/blob/main/analyze-mlflow-trace/SKILL.md --- ## When it is useful One MLflow trace ID — debug, RCA, or quality review. ## The Catch Twin-adjacent to retrieving (search vs one ID). Generic root-cause steal. --- # MLflow Querying Metrics > Aggregated token/latency/count from tracking server **Subcategory:** MLflow **License:** Apache-2.0 **Source:** https://github.com/mlflow/skills/blob/main/querying-mlflow-metrics/SKILL.md --- ## When it is useful Aggregated token/latency/count/quality metrics from an MLflow tracking server. ## The Catch Ships fetch_metrics.py beside SKILL.md. Distinct from logfire-query and axiom-query-metrics (MLflow tracking server, not Logfire OTel / APL). --- # W&B Primary > Mixed runs, artifacts, Weave, Reports, Launch **Subcategory:** Weights & Biases **License:** Apache-2.0 **Source:** https://github.com/wandb/skills/blob/main/skills/wandb-primary/SKILL.md --- ## When it is useful Mixed Weights & Biases work — runs, artifacts, Weave traces, Reports, Launch. ## The Catch Parent/always-on (what’s happening in this W&B project). Loud: Personal Free $0/mo for personal AI apps/models (up to 5 model seats, 5 GB/mo storage, 1 GB/mo Weave ingestion) but corporate use is not allowed. Do not invent Weave-seat numbers. Distinct from mlflow-onboarding. --- # W&B Autoresearch > Bounded training research via W&B Launch **Subcategory:** Weights & Biases **License:** Apache-2.0 **Source:** https://github.com/wandb/skills/blob/main/skills/wandb-autoresearch/SKILL.md --- ## When it is useful Bounded training research via W&B Launch (queues, serial trials, resumable state). ## The Catch Launch-only — must not substitute local wandb.init() for Launch compute. Vendor lock. Distinct from MLflow tracing. --- # W&B Eval Tables > Non-destructive EvalTable previews from wandb.Table **Subcategory:** Weights & Biases **License:** Apache-2.0 **Source:** https://github.com/wandb/skills/blob/main/skills/wandb-eval-tables/SKILL.md --- ## When it is useful Convert wandb.Table artifacts to non-destructive EvalTable previews. ## The Catch Creates new runs/keys; never overwrites source. Distinct from logfire-evals and HOLD mlflow agent-evaluation. --- # MinIO mc > AIStor mc CLI for S3-compatible storage **Subcategory:** MinIO **License:** Apache-2.0 **Source:** https://github.com/minio/skills/blob/main/mc/SKILL.md --- ## When it is useful AIStor mc CLI against S3-compatible storage (alias, ls/cp, versioning, replication, Tables). ## The Catch Targets AIStor — github.com/minio/minio is archived. Requires mc on PATH. Destructive ops. Works against AWS S3 too (false-fire vs AWS CLI). Distinct from neon-object-storage. AIStor Free $0 standalone/single-node only (no distributed HA). --- # MinIO AIMem > AIStor Memory FUSE workspace for a remote bucket **Subcategory:** MinIO **License:** Apache-2.0 **Source:** https://github.com/minio/skills/blob/main/aimem/SKILL.md --- ## When it is useful Inside an AIStor Memory FUSE workspace (POSIX reads/writes to a remote bucket). ## The Catch Never-fires unless that mount exists. Not generic agent memory. Targets AIStor — minio/minio archived. --- # MinIO Docs > ISO 24495-1 plain-language README/guides **Subcategory:** MinIO **License:** Apache-2.0 **Source:** https://github.com/minio/skills/blob/main/docs/SKILL.md --- ## When it is useful Writing README/guides to ISO 24495-1 Plain Language for MinIO/AIStor. ## The Catch Always-on docs steal (same class as wandb-primary). Kept with Catch so this is not a 2-drip. Targets AIStor — minio/minio archived. HOLD pr-autofix. --- # Tigris Object Operations > put/get/remove/list/head/presigned via @tigrisdata/storage **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-object-operations/SKILL.md --- ## When it is useful Put/get/remove/list/head/presigned objects with native @tigrisdata/storage. ## The Catch Vendor lock native @tigrisdata/storage not boto3. Distinct from minio-mc. False-fires on generic upload. Overlaps HOLD file-storage. Free 5GB Standard + 10k Class A + 100k Class B/mo, no CC, egress 0. Do not reprint PAYG. --- # Tigris Bucket Management > Create/list/inspect/delete with Tigris snapshot/fork fields **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-bucket-management/SKILL.md --- ## When it is useful Create/list/inspect/delete buckets (access, region, snapshot enable, fork source). ## The Catch enableSnapshot/sourceBucketSnapshot is Tigris-only, not S3 CreateBucket. Distinct from minio-mc mb. --- # Tigris S3 Migration > Shadow-bucket/drain from S3, GCS, or Azure Blob **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-s3-migration/SKILL.md --- ## When it is useful Shadow-bucket/drain migration from AWS S3, GCS, or Azure Blob into Tigris. ## The Catch Vendor steal vs rclone/mc mirror. Not a MinIO to AIStor path. Does not cover OSS MinIO EOL. --- # Tigris Snapshots Forking > Whole-bucket PITR and copy-on-write forks **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-snapshots-forking/SKILL.md --- ## When it is useful Whole-bucket point-in-time recovery and copy-on-write forks. ## The Catch Not S3 object versioning. Distinct from HOLD snapshots-recovery. False-fires on git/Neon/Postgres fork. --- # Tigris Agent Kit > tigrisdata agent-kit forks, TTL workspaces, checkpoints **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-agent-kit/SKILL.md --- ## When it is useful Use tigrisdata agent-kit for forks, TTL workspaces, checkpoints, and bucket-webhook. ## The Catch Not generic agent storage. Overlaps snapshots-forking compose layer. Distinct from minio-aimem. Extra package manager package. --- # Tigris Lifecycle Management > Expiration, TTL delete, Standard to IA/Archive **Subcategory:** Tigris **License:** MIT **Source:** https://github.com/tigrisdata/skills/blob/main/skills/tigris-lifecycle-management/SKILL.md --- ## When it is useful Object expiration, TTL delete, and Standard to IA/Archive transitions. ## The Catch Free allowance is Standard 5 GB only. IA 30-day / Archive 90-day min retention still bills if TTL fires early. Distinct from object-ops remove. Do not reprint PAYG rates. --- # YugabyteDB YSQL > SQL/schema on YSQL port 5433 with shard keys **Subcategory:** YugabyteDB **License:** Apache-2.0 **Source:** https://github.com/yugabyte/skills/blob/main/skills/ysql/SKILL.md --- ## When it is useful SQL/schema on YSQL (port 5433) — sharding, smart drivers, txn retries, PG migration. ## The Catch Vendor lock to YB, not vanilla Postgres. Distinct from neon-postgres (HASH shard keys, 5433 not 5432, no Neon branches). False-fires on generic CREATE TABLE. OSS self-host Apache-2.0 $0. Aeon Sandbox Free: 1 node / <=2 vCPU / 4 GB RAM / 10 GB / 15 connections; paused after 10 days inactivity, deleted after 15. Do not reprint Aeon Dedicated $/vCPU. --- # YugabyteDB YCQL > CQL schema on YCQL port 9042 **Subcategory:** YugabyteDB **License:** Apache-2.0 **Source:** https://github.com/yugabyte/skills/blob/main/skills/ycql/SKILL.md --- ## When it is useful CQL schema on YCQL (port 9042). ## The Catch Not Apache Cassandra. YCQL secondary indexes are global + ACID. Distinct from yugabyte-ysql. OSS self-host Apache-2.0 $0. --- # YugabyteDB RAG LangChain > Hybrid vector+FTS via LangChain PGVectorStore **Subcategory:** YugabyteDB **License:** Apache-2.0 **Source:** https://github.com/yugabyte/skills/blob/main/skills/yb-rag-langchain/SKILL.md --- ## When it is useful RAG / hybrid vector+FTS with LangChain PGVectorStore on YugabyteDB. ## The Catch Not Pinecone/Qdrant. ANN is ybhnsw; IVFFlat unsupported. PGVectorStore v2 not v1. HybridSearchConfig top_k defaults to 4. --- # YugabyteDB K8s Operator > YBUniverse CRDs / Helm on Kubernetes **Subcategory:** YugabyteDB **License:** Apache-2.0 **Source:** https://github.com/yugabyte/skills/blob/main/skills/yb-k8s-operator/SKILL.md --- ## When it is useful YBUniverse CRDs / Helm on Kubernetes for YugabyteDB. ## The Catch Loud: folder yb-k8s-operator, frontmatter name is operator. Body covers YBA Kubernetes Operator (commercial) and does not fully apply to the YugabyteDB OSS Kubernetes Operator. Will prescribe YBA CRDs the OSS operator does not have. --- # Redpanda Streaming > Produce/consume on Redpanda Kafka API :9092 **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/streaming/SKILL.md --- ## When it is useful Produce/consume, consumer groups, transactions, topic config, Kafka client compatibility on Redpanda (Kafka API :9092). ## The Catch Not Apache Kafka/MSK. Skills Apache-2.0; Redpanda Community is BSL 1.1 (free source-available; cannot offer as commercial streaming service; converts to Apache-2.0 four years after each merge). Self-host Community $0. Iceberg/Cloud Topics/Tiered Storage here are Enterprise-licensed, not the $0 path. Distinct from GetStream/LiveKit. --- # Redpanda rpk Topic > Create/alter/produce/consume via rpk topic **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/rpk-topic/SKILL.md --- ## When it is useful Create/alter/produce/consume topics via rpk topic. ## The Catch rpk, not kafka-topics.sh. Overlaps streaming (this is the CLI). Distinct from minio-mc. Community BSL self-host $0. --- # Redpanda rpk Transform > In-broker Wasm transforms via rpk transform **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/rpk-transform/SKILL.md --- ## When it is useful In-broker Wasm transforms (TinyGo/Rust/JS/TS) via rpk transform. ## The Catch Not Kafka Streams/Flink/Connect. data_transforms_enabled triggers a rolling restart. TinyGo default: no goroutines. No network/disk inside Wasm. Not on Serverless (BYOC/Dedicated only). --- # Redpanda Connect > Declarative YAML pipelines (ex-Benthos) via rpk connect **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/connect/SKILL.md --- ## When it is useful Declarative YAML pipelines (ex-Benthos) with rpk connect run / Bloblang. ## The Catch Community $0 BSL; enterprise connectors blocked without a license after the Connect 30-day trial. Not the CDC dump (HOLD connect-cdc-*). --- # Redpanda rpk Redpanda > Self-managed broker lifecycle via rpk redpanda **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/rpk-redpanda/SKILL.md --- ## When it is useful Self-managed broker lifecycle (rpk redpanda start/stop/tune, redpanda.yaml). ## The Catch Self-managed only — hidden on macOS/Windows rpk builds; does not apply to Cloud. This is the $0 Community BSL path. Distinct from cloud-serverless. --- # Redpanda Cloud Serverless > Provision Serverless via Control Plane API **Subcategory:** Redpanda **License:** Apache-2.0 **Source:** https://github.com/redpanda-data/skills/blob/main/skills/cloud-serverless/SKILL.md --- ## When it is useful Provision Serverless clusters via Control Plane API api.redpanda.com. ## The Catch Loud — not a $0 hobby. $100/30d trial, no CC, then PAYG; suspend at expiry, 7-day grace, then delete. Cloud = Enterprise. GCP Serverless is beta this pass. Do not reprint MB/s or PAYG rates. Distinct from rpk-redpanda Community path. --- # ScyllaDB Data Modeling > CQL schema, partition keys, query-first design **Subcategory:** ScyllaDB **License:** Apache-2.0 **Source:** https://github.com/scylladb/agent-skills/blob/main/skills/scylladb-data-modeling/SKILL.md --- ## When it is useful CQL schema / partition+clustering keys, query-first design, ALLOW FILTERING, hot partitions. ## The Catch Scylla CQL, not Yugabyte ycql and not vanilla Cassandra. False-fires on generic schema design. Distinct from AY ycql (no ACID global indexes story). Skills Apache-2.0; Enterprise is source-available (OSS AGPL 6.2.x frozen). --- # ScyllaDB Alternator > DynamoDB-compatible API via Alternator **Subcategory:** ScyllaDB **License:** Apache-2.0 **Source:** https://github.com/scylladb/agent-skills/blob/main/skills/scylladb-alternator/SKILL.md --- ## When it is useful DynamoDB-compatible Alternator (boto3/AWS SDK, endpoint_url, write isolation). ## The Catch Not AWS DynamoDB. No single-endpoint like real DynamoDB — load balancing is the trap. Alternator tables are ordinary Scylla tables. Distinct from data-modeling. --- # ScyllaDB Kubernetes > Operator / ScyllaCluster CRDs, CPU pinning, NVMe **Subcategory:** ScyllaDB **License:** Apache-2.0 **Source:** https://github.com/scylladb/agent-skills/blob/main/skills/scylladb-kubernetes/SKILL.md --- ## When it is useful ScyllaDB Operator / ScyllaCluster CRDs, CPU pinning, local NVMe. ## The Catch Operator is Apache; the DB it deploys is source-available Enterprise (or frozen AGPL 6.2). Distinct from yugabyte-k8s-operator. --- # ScyllaDB Vector Search > HNSW / ANN / RAG vectors on ScyllaDB Cloud **Subcategory:** ScyllaDB **License:** Apache-2.0 **Source:** https://github.com/scylladb/agent-skills/blob/main/skills/scylladb-vector-search/SKILL.md --- ## When it is useful HNSW / ANN / RAG vectors on ScyllaDB Cloud. ## The Catch Loud — Cloud only; not available in OSS or self-managed. Distinct from yugabyte-rag-langchain (YSQL pgvector/ybhnsw). --- # ScyllaDB Cloud Setup > Connect apps to ScyllaDB Cloud clusters **Subcategory:** ScyllaDB **License:** Apache-2.0 **Source:** https://github.com/scylladb/agent-skills/blob/main/skills/scylladb-cloud-setup/SKILL.md --- ## When it is useful Connect an app to ScyllaDB Cloud (drivers, credentials, IP allow list). ## The Catch Cloud cluster, not self-host. Will prescribe Free Trial + VPC peering (peering cannot be enabled later). One trial per account, not a $0 hobby forever. Self-host Enterprise free tier is separate: 10 TB disk + 50 vCPU per org, Never Customers only. --- # Buildkite Pipelines > Write/optimize .buildkite/pipeline.yml **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-pipelines/SKILL.md --- ## When it is useful Write/optimize .buildkite/pipeline.yml (steps, matrix, plugins, dynamic pipelines). ## The Catch Vendor lock to Buildkite YAML, not GHA. False-fires on generic write-a-pipeline. Distinct from n8n / Trigger.dev durable jobs. Free plan $0 verified (5 users, 10 concurrent, 2k Linux vCPU-min/mo). --- # Buildkite CLI > bk CLI — trigger/watch/cancel builds **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-cli/SKILL.md --- ## When it is useful bk CLI — trigger/watch/cancel builds, logs, artifacts, local agent. ## The Catch bk, not buildkite-agent (that is agent-runtime). Distinct from minio-mc / rpk. --- # Buildkite API > REST/GraphQL at api.buildkite.com **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-api/SKILL.md --- ## When it is useful REST/GraphQL at api.buildkite.com, webhooks, pagination. ## The Catch Not the CLI. GraphQL + REST both in one skill. Distinct from pipelines YAML. --- # Buildkite Agent Runtime > In-step buildkite-agent annotate/artifacts/OIDC **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-agent-runtime/SKILL.md --- ## When it is useful In-step buildkite-agent (annotate, artifacts, OIDC, locks, redaction, pipeline upload). ## The Catch Agent binary inside a job, not bk CLI. Description also matches generic upload artifacts — false-fire risk. --- # Buildkite Migration > Convert GHA/Jenkins/Circle via bk pipeline convert **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-migration/SKILL.md --- ## When it is useful Convert GHA/Jenkins/CircleCI/Bitbucket/GitLab CI via bk pipeline convert. ## The Catch Vendor steal — will prescribe Buildkite even when staying on GHA is the gap. Convert hits a public API, no account/token required. Distinct from tigris-s3-migration. --- # Buildkite Preflight > Run Buildkite CI against local tree before push **Subcategory:** Buildkite **License:** MIT **Source:** https://github.com/buildkite/skills/blob/main/skills/buildkite-preflight/SKILL.md --- ## When it is useful Run Buildkite CI against the local working tree before push. ## The Catch Local working-tree loop, not writing YAML. Still needs bk auth + push access to origin — not an offline runner. Distinct from pipelines. --- # GrowthBook Feature Flags > Create/toggle/target/roll out GrowthBook flags **Subcategory:** GrowthBook **License:** MIT **Source:** https://github.com/growthbook/skills/blob/main/skills/feature-flags/SKILL.md --- ## When it is useful Create/toggle/target/roll out GrowthBook flags (draft → review → publish, kill switch). v2 /api/v2/features. ## The Catch Not Mixpanel flags. Generic create-a-flag false-fire. Distinct from mixpanel-manage-feature-flags. Scheduled flags / safe rollouts / ramps are paid — Starter is toggle+rollout, not Pro/Enterprise lifecycle. Cloud Starter $0 (3 users) or self-host MIT. --- # GrowthBook Experiments > Design/launch/analyze GrowthBook A/B tests **Subcategory:** GrowthBook **License:** MIT **Source:** https://github.com/growthbook/skills/blob/main/skills/experiments/SKILL.md --- ## When it is useful Design/launch/analyze/stop GrowthBook A/B tests (SRM, chance-to-win, guardrails). ## The Catch Bandits are identified then sent to the UI — these workflows do not operate bandits (Pro). False-fires on generic run-an-A/B-test. Distinct from Mixpanel experiment HOLDs. --- # GrowthBook Analytics > Product Analytics explorations + fact-metric catalog **Subcategory:** GrowthBook **License:** MIT **Source:** https://github.com/growthbook/skills/blob/main/skills/analytics/SKILL.md --- ## When it is useful Product Analytics explorations + metric/fact-table catalog (/api/v1/product-analytics, fact-metrics). ## The Catch Not Mixpanel dashboards. Mixpanel/GA datasources cannot run explorations. Distinct from mixpanel-create-dashboard. --- # GrowthBook Setup > Configure GB_API_KEY for GrowthBook skills **Subcategory:** GrowthBook **License:** MIT **Source:** https://github.com/growthbook/skills/blob/main/skills/gb-setup/SKILL.md --- ## When it is useful Configure GB_API_KEY / optional GB_API_URL for self-host so feature-flags, experiments, and analytics can run. ## The Catch Writes ~/.config/growthbook/.env chmod 600 — credential helper, not deploying GrowthBook. Folder gb-setup. Keep so this is not a 3-drip with a missing on-ramp. --- # Descope Auth > Add Descope login (OTP, passkeys, OAuth, SSO, MFA) **Subcategory:** Descope **License:** MIT **Source:** https://github.com/descope/skills/blob/main/skills/descope-auth/SKILL.md --- ## When it is useful Add Descope login (OTP, magic link, passkeys, OAuth, SSO, MFA) — Next.js/React/backend router. ## The Catch Vendor lock. False-fires on generic add-auth. Hosted widget, not BYOS. Free Forever 7.5k MAU / 10 tenants / 3 SSO. SMS/voice via Descope connectors: 100/mo then sign-in breaks unless BYO. Distinct from Clerk/WorkOS. --- # Descope FGA Schema > Author/apply Descope FGA ReBAC/ABAC DSL **Subcategory:** Descope **License:** MIT **Source:** https://github.com/descope/skills/blob/main/skills/descope-fga-schema/SKILL.md --- ## When it is useful Author/apply Descope FGA (ReBAC/ABAC DSL). ## The Catch Hard-stops without Descope Management MCP (DryRunSchema / CreateFGASchema). Not generic set-up-roles. Distinct from descope-auth (identity vs authorization model). --- # Descope BYOS Builder > React Bring-Your-Own-Screen over Descope flow JSON **Subcategory:** Descope **License:** MIT **Source:** https://github.com/descope/skills/blob/main/skills/descope-byos-builder/SKILL.md --- ## When it is useful React Bring-Your-Own-Screen over exported Descope flow JSON (state.next(interactionId)). ## The Catch Not the hosted widget. Body says don't use when hosted already works. Distinct from descope-auth. --- # Descope Terraform > Terraform provider for Descope projects **Subcategory:** Descope **License:** MIT **Source:** https://github.com/descope/skills/blob/main/skills/descope-terraform/SKILL.md --- ## When it is useful Terraform provider for Descope projects (auth methods, roles, connectors, flows). ## The Catch Loud — body requires Paid Descope License (Pro+). Not on Free Forever. Kept with Catch so this is not a 3-drip. --- # Flux GitOps Knowledge > Flux concepts + schema-validated GitOps YAML **Subcategory:** Flux CD **License:** Apache-2.0 **Source:** https://github.com/fluxcd/agent-skills/blob/main/skills/gitops-knowledge/SKILL.md --- ## When it is useful Flux concepts + schema-validated YAML for HelmRelease, Kustomization, GitRepository, OCIRepository, ResourceSet, FluxInstance. ## The Catch Always-on Flux steal on any how-does-GitOps-work. Explicitly not repo audit and not live-cluster debug. Prefer Flux Operator / FluxInstance; do not prescribe legacy flux bootstrap. CNCF Apache-2.0 self-host $0. --- # Flux GitOps Repo Audit > Audit a local GitOps repo (schema, RBAC, secrets) **Subcategory:** Flux CD **License:** Apache-2.0 **Source:** https://github.com/fluxcd/agent-skills/blob/main/skills/gitops-repo-audit/SKILL.md --- ## When it is useful Audit a local GitOps repo (schema validate, deprecated APIs, RBAC/multi-tenancy/secrets) — not a live cluster. ## The Catch Local files only. False-fires on generic security-check-this-repo. Distinct from knowledge (generates YAML) and cluster-debug (live). --- # Flux GitOps Cluster Debug > Debug stuck Flux resources on a live cluster **Subcategory:** Flux CD **License:** Apache-2.0 **Source:** https://github.com/fluxcd/agent-skills/blob/main/skills/gitops-cluster-debug/SKILL.md --- ## When it is useful Stuck/not-ready Flux resources on a live cluster (status, controller logs, dependency chains). ## The Catch Requires Flux Operator MCP; will not work from repo files alone. Distinct from repo-audit. --- # SigNoz MCP Setup > Wire/repair SigNoz MCP (Cloud OAuth vs self-host) **Subcategory:** SigNoz **License:** MIT **Source:** https://github.com/SigNoz/agent-skills/blob/main/plugins/signoz/skills/signoz-mcp-setup/SKILL.md --- ## When it is useful Wire/repair SigNoz MCP (Cloud OAuth vs self-host /mcp). On-ramp, not install SigNoz. ## The Catch Not installing SigNoz itself. Distinct from mongodb-mcp-setup. Community Edition OSS self-host $0 (full APM/logs/traces); Cloud starts paid — $0 path is self-host. --- # SigNoz Investigating Alerts > RCA a firing alert via neighbor metrics/traces/logs **Subcategory:** SigNoz **License:** MIT **Source:** https://github.com/SigNoz/agent-skills/blob/main/plugins/signoz/skills/signoz-investigating-alerts/SKILL.md --- ## When it is useful RCA a firing alert (neighbor metrics/traces/logs). Read-only; not creating alerts. ## The Catch Read-only RCA, not creating alerts (HOLD creating-alerts — always-on steal). --- # SigNoz Writing ClickHouse Queries > Raw ClickHouse SQL for a SigNoz dashboard panel **Subcategory:** SigNoz **License:** MIT **Source:** https://github.com/SigNoz/agent-skills/blob/main/plugins/signoz/skills/signoz-writing-clickhouse-queries/SKILL.md --- ## When it is useful Raw ClickHouse SQL for a SigNoz dashboard panel the builder cannot express. ## The Catch SigNoz panel SQL, not generic ClickHouse. Distinct from chdb-sql / infra-clickhouse. --- # SigNoz Reducing Telemetry Cost > Cost Meter, cardinality, drop volume **Subcategory:** SigNoz **License:** MIT **Source:** https://github.com/SigNoz/agent-skills/blob/main/plugins/signoz/skills/signoz-reducing-telemetry-cost/SKILL.md --- ## When it is useful Cost Meter, cardinality, drop volume. ## The Catch Loud: Cost Meter is Cloud-bill; Community self-host has no SigNoz invoice. Do not reprint per-GB rates. --- # SigNoz Managing Views > Explorer saved views (traces/logs/metrics/meter) **Subcategory:** SigNoz **License:** MIT **Source:** https://github.com/SigNoz/agent-skills/blob/main/plugins/signoz/skills/signoz-managing-views/SKILL.md --- ## When it is useful Explorer saved views (traces/logs/metrics/meter). ## The Catch Saved views, not dashboards. Distinct from creating-dashboards (HOLD — always-on monitoring steal). --- # Packer AWS AMI Builder > amazon-ebs builder for custom EC2 AMIs **Subcategory:** Packer **License:** MPL-2.0 **Source:** https://github.com/hashicorp/agent-skills/blob/main/plugins/packer/skills/aws-ami-builder/SKILL.md --- ## When it is useful amazon-ebs builder, custom EC2 AMIs. ## The Catch Frontmatter aws-ami-builder. Packer ≥1.10 is BSL 1.1 (CLI self-use $0; cannot offer competing hosted Packer). AWS bill for the build, not Packer license. Distinct from pulumi-component. HOLD Terraform 16 from same repo. --- # Packer Azure Image Builder > azure-arm builder for managed images / Gallery **Subcategory:** Packer **License:** MPL-2.0 **Source:** https://github.com/hashicorp/agent-skills/blob/main/plugins/packer/skills/azure-image-builder/SKILL.md --- ## When it is useful azure-arm builder, managed images / Compute Gallery. ## The Catch Azure bill for the build. Distinct from aws-ami-builder. --- # Packer Windows Builder > WinRM + PowerShell provisioners (AMI/Azure/VMware) **Subcategory:** Packer **License:** MPL-2.0 **Source:** https://github.com/hashicorp/agent-skills/blob/main/plugins/packer/skills/windows-builder/SKILL.md --- ## When it is useful WinRM + PowerShell provisioners (AMI/Azure/VMware). ## The Catch Communicator/provisioner, not a third cloud builder. Long builds; failed runs leave resources. --- # Packer Push to Registry > Push metadata (not images) to HCP Packer **Subcategory:** Packer **License:** MPL-2.0 **Source:** https://github.com/hashicorp/agent-skills/blob/main/plugins/packer/skills/push-to-registry/SKILL.md --- ## When it is useful Push metadata (not images) to HCP Packer. ## The Catch Loud — HCP Packer Essentials vs paid Standard; metadata-only, not a local file registry. Do not treat Standard as $0. Kept so this is not a 3-drip. --- # Elastic Observability K8s Investigation > CrashLoopBackOff / OOMKilled via EDOT OTel **Subcategory:** Elastic **License:** Apache-2.0 **Source:** https://github.com/elastic/agent-skills/blob/main/skills/observability/k8s-investigation/SKILL.md --- ## When it is useful CrashLoopBackOff / OOMKilled via EDOT OTel, not legacy ECS k8s. ES 8.11+ or Serverless; VALUES() GA only 9.4 on Stack. ## The Catch Elastic skills, not OpenSearch. Product is ELv2/SSPL (not OSI). Distinct from signoz-investigating-alerts. Cloud is 14-day trial then PAYG; self-managed Basic free under ELv2. --- # Elastic Security Detection Rule Management > SIEM/Endpoint detection rules via Kibana API **Subcategory:** Elastic **License:** Apache-2.0 **Source:** https://github.com/elastic/agent-skills/blob/main/skills/security/detection-rule-management/SKILL.md --- ## When it is useful SIEM/Endpoint detection rules, exceptions, noisy rules via Kibana API. ## The Catch Elastic Security, not generic SIEM. Not OpenSearch. --- # Elastic Security Alert Triage > Triage Elastic Security alerts, classify, acknowledge **Subcategory:** Elastic **License:** Apache-2.0 **Source:** https://github.com/elastic/agent-skills/blob/main/skills/security/alert-triage/SKILL.md --- ## When it is useful Triage Elastic Security alerts, classify, acknowledge. ## The Catch Depends on case-management. Distinct from signoz-investigating-alerts. --- # Elastic Security Case Management > Kibana Cases API (securitySolution only) **Subcategory:** Elastic **License:** Apache-2.0 **Source:** https://github.com/elastic/agent-skills/blob/main/skills/security/case-management/SKILL.md --- ## When it is useful Kibana Cases API, securitySolution only. ## The Catch Kept so alert-triage is not orphaned. Not generic ticketing. --- # Elastic Cloud Provisioning > Create/update/delete Serverless + Hosted deployments **Subcategory:** Elastic **License:** Apache-2.0 **Source:** https://github.com/elastic/agent-skills/blob/main/skills/cloud/provisioning/SKILL.md --- ## When it is useful Create/update/delete Serverless projects + Hosted deployments, traffic filters. ## The Catch Elastic Cloud API only (not self-managed). Depends on HOLD cloud-onboarding for CLI context. Cloud is 14-day trial (no CC) then PAYG — not a $0 forever hatch. Do not reprint VCU/Hosted rates. --- # Grafana Alloy > Unified OTel-compatible collector (config.alloy) **Subcategory:** Grafana **License:** Apache-2.0 **Source:** https://github.com/grafana/skills/blob/main/skills/grafana-core/alloy/SKILL.md --- ## When it is useful Unified OTel-compatible collector (metrics/logs/traces/profiles, config.alloy). ## The Catch Frontmatter alloy. Steals generic set-up-the-agent / OTel collector config. OSS CLI. Distinct from generic OTel. Grafana Cloud Free always $0 (10k series / 50GB logs/traces/profiles / 14d / 3 users). --- # Grafana Beyla > eBPF zero-code HTTP/gRPC/DB auto-instrument **Subcategory:** Grafana **License:** Apache-2.0 **Source:** https://github.com/grafana/skills/blob/main/skills/grafana-core/beyla/SKILL.md --- ## When it is useful eBPF zero-code HTTP/gRPC/DB auto-instrument (Linux 5.8+ BTF, CAP_SYS_ADMIN). ## The Catch Frontmatter beyla. Kernel/privileged host; not an SDK. Distinct from OTel SDKs. --- # Grafana Loki > Log aggregation + LogQL **Subcategory:** Grafana **License:** Apache-2.0 **Source:** https://github.com/grafana/skills/blob/main/skills/grafana-lgtm/loki/SKILL.md --- ## When it is useful Log aggregation + LogQL. ## The Catch Frontmatter loki. Indexes labels not full text (not Elastic/OpenSearch). Distinct from Elastic BN. OSS self-host AGPL $0; Cloud Free 50GB ingest / 14d. --- # Grafana Tempo > Tracing backend (OTLP/Jaeger/Zipkin, TraceQL) **Subcategory:** Grafana **License:** Apache-2.0 **Source:** https://github.com/grafana/skills/blob/main/skills/grafana-lgtm/tempo/SKILL.md --- ## When it is useful Tracing backend (OTLP/Jaeger/Zipkin, TraceQL, object storage). ## The Catch Frontmatter tempo. Steals generic tracing-backend / find-slow-requests. Distinct from SigNoz and from live temporal-* (workflow engine). --- # Grafana Pyroscope > Continuous profiling / flame graphs **Subcategory:** Grafana **License:** Apache-2.0 **Source:** https://github.com/grafana/skills/blob/main/skills/grafana-lgtm/pyroscope/SKILL.md --- ## When it is useful Continuous profiling / flame graphs (SDK push or Alloy eBPF). ## The Catch Frontmatter pyroscope. eBPF path needs Linux 5.8+ BTF. Distinct from traces/logs. Cloud Free 50GB profiles / 14d. --- # Render Blueprints > Author/validate render.yaml IaC **Subcategory:** Render **License:** MIT **Source:** https://github.com/render-oss/skills/blob/main/skills/render-blueprints/SKILL.md --- ## When it is useful Author/validate render.yaml IaC (fromDatabase/fromService, previews). ## The Catch Blueprint sync needs GitHub/GitLab/Bitbucket. CLI v2.7.0+ for validate. Proprietary PaaS; Hobby $0 path exists. Distinct from Railway. --- # Render Postgres > Render Managed Postgres **Subcategory:** Render **License:** MIT **Source:** https://github.com/render-oss/skills/blob/main/skills/render-postgres/SKILL.md --- ## When it is useful Render Managed Postgres. ## The Catch Loud — Free instance is a 30-day limit, not $0 forever. Steals generic Postgres/DATABASE_URL. Distinct from Neon/Yugabyte. Do not reprint Basic/Pro rates. --- # Render Key Value > Redis-compatible Valkey 8 on Render **Subcategory:** Render **License:** MIT **Source:** https://github.com/render-oss/skills/blob/main/skills/render-keyvalue/SKILL.md --- ## When it is useful Redis-compatible Valkey 8. ## The Catch Free is 25MB, no persistence. Steals generic Redis/cache. Distinct from Redis/Upstash. --- # Render CLI > Deploys, logs, SSH, psql, Blueprint validate **Subcategory:** Render **License:** MIT **Source:** https://github.com/render-oss/skills/blob/main/skills/render-cli/SKILL.md --- ## When it is useful Deploys, logs, SSH, psql, Blueprint validate, RENDER_API_KEY. ## The Catch CLI v2.7.0+; not a substitute for HOLD render-deploy (twin of live openai render-deploy). --- # Render MCP > Hosted MCP at mcp.render.com **Subcategory:** Render **License:** MIT **Source:** https://github.com/render-oss/skills/blob/main/skills/render-mcp/SKILL.md --- ## When it is useful Hosted MCP at mcp.render.com (Cursor/Claude/Codex). ## The Catch On-ramp; keep so this is not a 3-drip. Distinct from Railway MCP. --- # Confluent Kafka Schema Registry > Scan repo, extract schemas, Terraform for Schema Registry **Subcategory:** Confluent **License:** Apache-2.0 **Source:** https://github.com/confluentinc/agent-skills/blob/main/skills/kafka-schema-registry/SKILL.md --- ## When it is useful Scan a repo for Kafka usage, extract schemas, generate Terraform for Confluent Schema Registry. ## The Catch Frontmatter kafka-schema-registry. Distinct from generic Avro. Distinct from Redpanda AZ (Kafka-compatible broker BSL). Cloud Basic $0/mo first eCKU free; OSS Kafka self-host $0. --- # Confluent Kafka Streams > JVM-embedded Kafka Streams (KStream/KTable) **Subcategory:** Confluent **License:** Apache-2.0 **Source:** https://github.com/confluentinc/agent-skills/blob/main/skills/kafka-streams-programming/SKILL.md --- ## When it is useful JVM-embedded Kafka Streams (KStream/KTable, TopologyTestDriver). ## The Catch Frontmatter kafka-streams-programming. JVM-only; also triggers on WarpStream tuning. Distinct from Redpanda / Flink. --- # Confluent Kafka Java Client > Java producer/consumer + Schema Registry serializers **Subcategory:** Confluent **License:** Apache-2.0 **Source:** https://github.com/confluentinc/agent-skills/blob/main/skills/developing-kafka-java-client/SKILL.md --- ## When it is useful Apache Kafka Java producer/consumer + Schema Registry serializers; Cloud, local Docker, or WarpStream. ## The Catch Frontmatter developing-kafka-java-client. Local Docker is the $0 OSS path. HOLD the Python sibling as a twin. --- # Confluent Flink UDF > Java Flink UDFs/UDTFs/PTFs for Cloud or local Docker **Subcategory:** Confluent **License:** Apache-2.0 **Source:** https://github.com/confluentinc/agent-skills/blob/main/skills/flink-udf/SKILL.md --- ## When it is useful Java Flink UDFs/UDTFs/PTFs for Cloud or local Docker. ## The Catch Frontmatter flink-udf. Java only; not general Flink jobs. --- # Confluent Cloud Flink SQL > Flink SQL on Confluent Cloud (CC ≠ OSS dialect) **Subcategory:** Confluent **License:** Apache-2.0 **Source:** https://github.com/confluentinc/agent-skills/blob/main/skills/confluent-cloud-flink-sql/SKILL.md --- ## When it is useful Flink SQL on Confluent Cloud (CC ≠ OSS Flink dialect). ## The Catch Requires authenticated confluent CLI + a Cloud compute pool; statement runs consume CFUs. Keep with loud Catch so this is not a 4-drip. Do not reprint CFU rates. --- # PostHog Instrument Error Tracking > Capture exceptions + source maps **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/instrument-error-tracking/SKILL.md --- ## When it is useful Capture exceptions + source maps. ## The Catch Frontmatter instrument-error-tracking. Also does initial SDK setup (on-ramp-adjacent). Distinct from Sentry / Firebase Crashlytics. Cloud Free forever 100K exceptions/mo; OSS self-host $0. --- # PostHog Creating Experiments > 3-step experiment create (hypothesis → rollout → analytics) **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/creating-experiments/SKILL.md --- ## When it is useful 3-step experiment create (hypothesis → rollout → analytics). ## The Catch Frontmatter creating-experiments. Steals generic create-an-A/B-test. Distinct from GrowthBook. Do not reprint paid flag overage. --- # PostHog Exploring LLM Traces > Debug LLM/agent traces (posthog.ai_events) **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/exploring-llm-traces/SKILL.md --- ## When it is useful Debug LLM/agent traces (posthog.ai_events, not events.properties.$ai_*). ## The Catch Frontmatter exploring-llm-traces. MCP/HogQL-specific. Cloud Free 100K AI events. Distinct from already-live instrument-llm-analytics (HOLD as twin). --- # PostHog Investigating Replay > One-pass session recording + person + linked errors **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/investigating-replay/SKILL.md --- ## When it is useful One-pass session recording + person + linked errors. ## The Catch Frontmatter investigating-replay. Needs a recording/session ID, not watch-recordings. Cloud Free 5K recordings/mo. --- # PostHog Cleaning Up Stale Feature Flags > Audit unused/fully-rolled-out flags and safe delete **Subcategory:** PostHog **License:** MIT **Source:** https://github.com/PostHog/skills/blob/main/skills/omnibus/cleaning-up-stale-feature-flags/SKILL.md --- ## When it is useful Audit unused/fully-rolled-out flags and safe delete. ## The Catch Frontmatter cleaning-up-stale-feature-flags. Distinct from Mixpanel flags / GrowthBook. Cloud Free 1M flag requests/mo. --- # Letta API Client > Python/TS SDK for stateful Letta API **Subcategory:** Letta **License:** MIT **Source:** https://github.com/letta-ai/skills/blob/main/letta/letta-api-client/SKILL.md --- ## When it is useful Python/TS SDK for the stateful Letta API (agents, memory, tools, multi-user). ## The Catch API Plan is usage-based; Cloud Free personal is 3 agents + BYOK. OSS self-host $0 (no Letta account). Do not reprint CPU-sec rates. --- # Letta Configuration > Model handles, BYOK providers, self-hosted env vars **Subcategory:** Letta **License:** MIT **Source:** https://github.com/letta-ai/skills/blob/main/letta/letta-configuration/SKILL.md --- ## When it is useful Model handles, BYOK providers, self-hosted env vars. ## The Catch Cloud Free is BYOK/rotating free models, not unlimited hosted inference. Self-host docs: docs.letta.com/self-hosting. --- # Letta Fleet Management > kubectl-style lettactl for multi-agent fleets **Subcategory:** Letta **License:** MIT **Source:** https://github.com/letta-ai/skills/blob/main/letta/fleet-management/SKILL.md --- ## When it is useful kubectl-style lettactl for multi-agent fleets (canary, multi-tenancy). ## The Catch Frontmatter already letta-fleet-management. Fleet ops vs single-agent API. --- # Letta Code Channels > Letta Code channel adapters (Telegram/WhatsApp/Slack/Discord) **Subcategory:** Letta **License:** MIT **Source:** https://github.com/letta-ai/skills/blob/main/letta/creating-letta-code-channels/SKILL.md --- ## When it is useful Letta Code channel adapters (~/.letta/channels: Telegram/WhatsApp/Slack/Discord). ## The Catch Frontmatter creating-letta-code-channels — prefixed. Letta Code surface, not the API. --- # Letta Filesystem to MemFS > Migrate deprecated Filesystem folders to MemFS **Subcategory:** Letta **License:** MIT **Source:** https://github.com/letta-ai/skills/blob/main/letta/letta-filesystem-to-memfs/SKILL.md --- ## When it is useful Migrate deprecated Filesystem folders to MemFS. ## The Catch Migration-only; not a generic file-search steal. --- # Firecrawl Scrape > CLI: extract a URL as clean markdown (JS-rendered OK) **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-scrape/SKILL.md --- ## When it is useful Extract a URL as clean markdown, including JS-rendered pages. Prefer over WebFetch. ## The Catch CLI one-URL scrape. Distinct from live firecrawl-build-scrape (app /scrape integration). Product AGPL-3.0 self-host $0; Cloud Free 1,000 credits/mo, no card, no rollover. --- # Firecrawl Search > CLI: web search with optional full page content **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-search/SKILL.md --- ## When it is useful Web search with optional full page content when no URL is known. ## The Catch CLI search. Distinct from live firecrawl-build-search. Feedback refund is a 2-minute window — not a free-forever extra. --- # Firecrawl Interact > CLI: drive a live browser on a scraped page **Subcategory:** Firecrawl **License:** ISC **Source:** https://github.com/firecrawl/skills/blob/main/skills/core/firecrawl-interact/SKILL.md --- ## When it is useful Drive a live browser on a scraped page (click, fill, paginate). Scrape first. ## The Catch CLI session. Distinct from live firecrawl-build-interact. Overlaps Playwright; this is Firecrawl-hosted browser minutes (2 credits/min). --- # Apollo Client > Apollo Client 4.x for React — queries, cache, local state **Subcategory:** Apollo GraphQL **License:** MIT **Source:** https://github.com/apollographql/skills/blob/main/skills/apollo-client/SKILL.md --- ## When it is useful Apollo Client 4.x for React — queries, mutations, cache, local state. ## The Catch React 18+/19. Distinct from iOS/Kotlin twins (HOLD). GraphOS Free forever, no CC, 3 users; OSS Client/Server self-host $0. --- # Apollo Server > Apollo Server 5.x — schemas, resolvers, auth, plugins **Subcategory:** Apollo GraphQL **License:** MIT **Source:** https://github.com/apollographql/skills/blob/main/skills/apollo-server/SKILL.md --- ## When it is useful Apollo Server 5.x — schemas, resolvers, auth, plugins. ## The Catch OSS self-host $0. Not GraphOS. --- # Apollo Federation > Federation 2.x subgraph schemas (@key, @shareable) **Subcategory:** Apollo GraphQL **License:** MIT **Source:** https://github.com/apollographql/skills/blob/main/skills/apollo-federation/SKILL.md --- ## When it is useful Federation 2.x subgraph schemas (@key, @shareable, composition). ## The Catch Schema authoring, not the Router runtime. --- # Apollo Router > Configure/run Apollo Router for federated supergraphs **Subcategory:** Apollo GraphQL **License:** MIT **Source:** https://github.com/apollographql/skills/blob/main/skills/apollo-router/SKILL.md --- ## When it is useful Configure/run Apollo Router for federated supergraphs (v1/v2 YAML). ## The Catch Loud — GraphOS Free Router is rate-limited 60 req/min. Coprocessors are not Free. --- # Apollo Rover > Rover CLI — publish/fetch/compose/check schemas **Subcategory:** Apollo GraphQL **License:** MIT **Source:** https://github.com/apollographql/skills/blob/main/skills/rover/SKILL.md --- ## When it is useful Rover CLI — publish/fetch/compose/check schemas, rover dev. ## The Catch Frontmatter name is rover — prefixed to apollo-rover. --- # Camunda c8ctl > Install/configure c8ctl, local cluster via c8run **Subcategory:** Camunda **License:** Apache-2.0 **Source:** https://github.com/camunda/skills/blob/main/skills/camunda-c8ctl/SKILL.md --- ## When it is useful Install/configure c8ctl, local cluster via c8run, profiles. ## The Catch Foundation for other camunda-* skills. Loud — $0 is Self-Managed Development local/non-prod only; production Self-Managed needs Enterprise. Engine is Camunda License 1.0 (not OSI). Distinct from Temporal (MIT) and n8n. --- # Camunda BPMN > Create/edit/validate BPMN 2.0 for Camunda 8 (Zeebe) **Subcategory:** Camunda **License:** Apache-2.0 **Source:** https://github.com/camunda/skills/blob/main/skills/camunda-bpmn/SKILL.md --- ## When it is useful Create/edit/validate BPMN 2.0 for Camunda 8 (Zeebe). ## The Catch Executable Zeebe BPMN, not generic BPMN. --- # Camunda FEEL > Write/debug FEEL expressions in BPMN, DMN, Forms **Subcategory:** Camunda **License:** Apache-2.0 **Source:** https://github.com/camunda/skills/blob/main/skills/camunda-feel/SKILL.md --- ## When it is useful Write/debug FEEL expressions used in BPMN, DMN, and Forms. ## The Catch Camunda FEEL, not a generic expression steal. --- # Camunda DMN > Author DMN decision tables and literal expressions **Subcategory:** Camunda **License:** Apache-2.0 **Source:** https://github.com/camunda/skills/blob/main/skills/camunda-dmn/SKILL.md --- ## When it is useful Author DMN decision tables and literal expressions. ## The Catch Camunda 8 DMN, not a generic rules dump. --- # Camunda Job Workers > Job workers in Java, Spring Boot, or TypeScript **Subcategory:** Camunda **License:** Apache-2.0 **Source:** https://github.com/camunda/skills/blob/main/skills/camunda-job-workers/SKILL.md --- ## When it is useful Implement job workers in Java, Spring Boot, or TypeScript. ## The Catch Workers against Zeebe service tasks. --- # ImageKit Upload Files > Upload images/videos from a public URL into the library **Subcategory:** ImageKit **License:** MIT **Source:** https://github.com/imagekit-developer/skills/blob/main/skills/upload-files/SKILL.md --- ## When it is useful Upload images/videos from a public URL into the media library. ## The Catch Frontmatter upload-files. Local paths unsupported — URL only. Proprietary DAM/CDN; Forever Free $0/mo (20 GB bandwidth + 3 GB DAM + 2 users; delivery stops at the cap). Distinct from Tigris. --- # ImageKit Transformation Builder > Image/video transformation URLs (resize, crop, AI, overlays) **Subcategory:** ImageKit **License:** MIT **Source:** https://github.com/imagekit-developer/skills/blob/main/skills/transformation-builder/SKILL.md --- ## When it is useful Image/video transformation URLs (resize, crop, AI edits, overlays). ## The Catch Frontmatter transformation-builder. Forever Free AI ops are limited. --- # ImageKit Search Assets > Lucene-like searchQuery for files/folders **Subcategory:** ImageKit **License:** MIT **Source:** https://github.com/imagekit-developer/skills/blob/main/skills/search-assets/SKILL.md --- ## When it is useful Lucene-like searchQuery for files/folders. ## The Catch Frontmatter search-assets. DAM search, not generic file search. --- # ImageKit AI Tasks > AI tagging, metadata extraction, quality checks **Subcategory:** ImageKit **License:** MIT **Source:** https://github.com/imagekit-developer/skills/blob/main/skills/ai-tasks/SKILL.md --- ## When it is useful AI tagging, metadata extraction, quality checks with controlled vocabularies. ## The Catch Frontmatter ai-tasks. Forever Free AI is capped — do not treat as unlimited. --- # ImageKit Integrations > Index of SDKs/plugins (front-end, back-end, mobile, CMS) **Subcategory:** ImageKit **License:** MIT **Source:** https://github.com/imagekit-developer/skills/blob/main/skills/imagekit-integrations/SKILL.md --- ## When it is useful Index of SDKs/plugins (front-end, back-end, mobile, CMS, storage). ## The Catch Kept so this is not a 4-drip. Index, not implementation (that is HOLD search-docs). --- # ZenML Quick Wins > Investigate stack, recommend, implement high-impact ZenML features **Subcategory:** ZenML **License:** MIT **Source:** https://github.com/zenml-io/skills/blob/main/skills/zenml-quick-wins/skills/quick-wins/SKILL.md --- ## When it is useful Improve a ZenML setup — metadata logging, experiment tracking, alerts, schedules/triggers, streaming events, secrets, Model Control Plane, or other ~5-minute MLOps quick wins. ## The Catch Loud: OSS self-host Free forever is the $0 path; Hosted Cloud/Pro paid (Cloud 14-day trial then paid). Do not reprint Pro $399. Distinct from MLflow and W&B. Skills MIT; product ZenML Apache-2.0. HOLD nine migration twins (argo/azureml/dagster/databricks/flyte/kedro/metaflow/prefect/vertexai). --- # ZenML Pipeline Authoring > Author @step/@pipeline, artifacts, Docker, materializers, YAML **Subcategory:** ZenML **License:** MIT **Source:** https://github.com/zenml-io/skills/blob/main/skills/zenml-pipeline-authoring/skills/pipeline-authoring/SKILL.md --- ## When it is useful Write ZenML pipelines/steps, Docker/Resource settings, materializers, metadata, secrets, deployments, streaming events, YAML config, or visualizations — including works locally but fails on cloud. ## The Catch Loud: OSS self-host Free forever is the $0 path; Hosted Cloud/Pro paid (Cloud 14-day trial then paid). Do not reprint Pro $399. Distinct from MLflow and W&B. Pair with zenml-scoping for multi-pipeline ambition. --- # ZenML Scoping > Decompose ambitious ML workflows into pipeline architectures **Subcategory:** ZenML **License:** MIT **Source:** https://github.com/zenml-io/skills/blob/main/skills/zenml-scoping/skills/scoping/SKILL.md --- ## When it is useful Scope complex or over-engineered MLOps plans into realistic multi-pipeline ZenML architectures before authoring code; produce pipeline_architecture.md. ## The Catch Loud: OSS self-host Free forever is the $0 path; Hosted Cloud/Pro paid (Cloud 14-day trial then paid). Do not reprint Pro $399. Distinct from MLflow and W&B. Architecture interview, not a mega-pipeline dump. --- # ZenML Airflow Migration > Migrate Airflow DAGs to idiomatic ZenML (frontmatter airflow-to-zenml-migration) **Subcategory:** ZenML **License:** MIT **Source:** https://github.com/zenml-io/skills/blob/main/skills/zenml-airflow-migration/skills/airflow-migration/SKILL.md --- ## When it is useful Migrate Apache Airflow DAGs/operators/XCom/sensors to idiomatic ZenML — concept map, code translation, severity-classified unsupported patterns. ## The Catch Frontmatter name is airflow-to-zenml-migration. Loud: OSS self-host Free forever is the $0 path; Hosted Cloud/Pro paid (Cloud 14-day trial then paid). Do not reprint Pro $399. Distinct from MLflow and W&B. HOLD other platform migration twins. --- # ZenML SageMaker Migration > Migrate SageMaker Pipelines DSL to idiomatic ZenML **Subcategory:** ZenML **License:** MIT **Source:** https://github.com/zenml-io/skills/blob/main/skills/zenml-sagemaker-migration/skills/sagemaker-migration/SKILL.md --- ## When it is useful Migrate Amazon SageMaker Pipelines (ProcessingStep/TrainingStep/PropertyFile/JsonGet) to idiomatic ZenML pipelines with concept mapping and unsupported-pattern flags. ## The Catch Frontmatter name is sagemaker-to-zenml-migration. Loud: OSS self-host Free forever is the $0 path; Hosted Cloud/Pro paid (Cloud 14-day trial then paid). Do not reprint Pro $399. Distinct from MLflow and W&B. HOLD other platform migration twins. --- # LaunchDarkly Flag Create > Create/configure flags to match existing codebase patterns **Subcategory:** LaunchDarkly **License:** Apache-2.0 **Source:** https://github.com/launchdarkly/agent-skills/blob/main/skills/feature-flags/launchdarkly-flag-create/SKILL.md --- ## When it is useful Create a new LaunchDarkly feature flag, wrap code in a flag, add a feature toggle, or set up flag scaffolding for an experiment — explore existing patterns first. ## The Catch Loud: Developer Free forever $0 no CC (unlimited seats/flags; 5 service connections + 1K client-side MAU/mo; Free ingestion stops at caps). Foundation PAYG after 14-day trial. Not OSS. Distinct from GrowthBook (OSS self-host) and PostHog experiments. Skills Apache-2.0; product proprietary SaaS. Requires LaunchDarkly MCP. --- # LaunchDarkly Flag Targeting > Toggle, percentage rollouts, rules, individual targets, env copy **Subcategory:** LaunchDarkly **License:** Apache-2.0 **Source:** https://github.com/launchdarkly/agent-skills/blob/main/skills/feature-flags/launchdarkly-flag-targeting/SKILL.md --- ## When it is useful Change who sees a LaunchDarkly flag — on/off, percentage rollouts, targeting rules, individual targets, or copy config between environments. ## The Catch Loud: Developer Free forever $0 no CC (unlimited seats/flags; 5 service connections + 1K client-side MAU/mo; Free ingestion stops at caps). Foundation PAYG after 14-day trial. Not OSS. Distinct from GrowthBook (OSS self-host) and PostHog experiments. Skills Apache-2.0; product proprietary SaaS. Requires LaunchDarkly MCP. --- # LaunchDarkly Guarded Rollout > Progressive rollout with metric monitoring and auto rollback **Subcategory:** LaunchDarkly **License:** Apache-2.0 **Source:** https://github.com/launchdarkly/agent-skills/blob/main/skills/feature-flags/launchdarkly-guarded-rollout/SKILL.md --- ## When it is useful Configure a guarded rollout — progressive traffic stages, monitored metrics, regression thresholds, automatic pause/rollback. ## The Catch Loud: Developer Free forever $0 no CC (unlimited seats/flags; 5 service connections + 1K client-side MAU/mo; Free ingestion stops at caps). Foundation PAYG after 14-day trial. Not OSS. Distinct from GrowthBook (OSS self-host) and PostHog experiments. Skills Apache-2.0; product proprietary SaaS. Requires LaunchDarkly MCP. Guarded rollout may climb into Foundation/Guardian paid path — do not invent it as Free. --- # LaunchDarkly Flag Cleanup > Safely remove a flag from code preserving production behavior **Subcategory:** LaunchDarkly **License:** Apache-2.0 **Source:** https://github.com/launchdarkly/agent-skills/blob/main/skills/feature-flags/launchdarkly-flag-cleanup/SKILL.md --- ## When it is useful Remove a LaunchDarkly flag from code after rollout — hardcode the winning variation, verify readiness, archive/delete carefully. ## The Catch Loud: Developer Free forever $0 no CC (unlimited seats/flags; 5 service connections + 1K client-side MAU/mo; Free ingestion stops at caps). Foundation PAYG after 14-day trial. Not OSS. Distinct from GrowthBook (OSS self-host) and PostHog experiments. Skills Apache-2.0; product proprietary SaaS. Requires LaunchDarkly MCP. HOLD flag-discovery twin. --- # LaunchDarkly Experiment Setup > Create experiments with metrics, treatments, start/stop iterations **Subcategory:** LaunchDarkly **License:** Apache-2.0 **Source:** https://github.com/launchdarkly/agent-skills/blob/main/skills/experiments/launchdarkly-experiment-setup/SKILL.md --- ## When it is useful Set up and run LaunchDarkly experiments — metrics, treatments, flag config, start iterations, swap design, stop with a winner. ## The Catch Loud: Developer Free forever $0 no CC (unlimited seats/flags; 5 service connections + 1K client-side MAU/mo; Free ingestion stops at caps). Foundation PAYG after 14-day trial. Not OSS. Distinct from GrowthBook (OSS self-host) and PostHog experiments. Skills Apache-2.0; product proprietary SaaS. Requires LaunchDarkly MCP. --- # Builder Visual Edit > Inspect/edit a running local app visually in Design iframes **Subcategory:** Builder.io **License:** MIT **Source:** https://github.com/BuilderIO/skills/blob/main/skills/visual-edit/SKILL.md --- ## When it is useful Inspect, compare, or edit a real local app visually in Design overview — URL-backed iframe screens, flow review, duplication, route-state exploration. ## The Catch Loud: Fusion Free forever $0 — 15 daily / 60 monthly Agent Credits; stops at cap; no rollover; proprietary SaaS (not OSS). Distinct from Expo leftovers / Firecrawl CLI. Skills MIT; product proprietary Fusion SaaS. --- # Builder Visual Plan > Turn text plans into interactive visual plans with diagrams **Subcategory:** Builder.io **License:** MIT **Source:** https://github.com/BuilderIO/skills/blob/main/skills/visual-plan/SKILL.md --- ## When it is useful Turn ordinary text plans into rich interactive visual plans with diagrams, file maps, annotated code, open questions, and UI/prototype review. ## The Catch Loud: Fusion Free forever $0 — 15 daily / 60 monthly Agent Credits; stops at cap; no rollover; proprietary SaaS (not OSS). Distinct from Expo leftovers / Firecrawl CLI. Skills MIT; product proprietary Fusion SaaS. --- # Builder Visual Recap > Visual recap of a PR/branch/commit/diff with diagrams **Subcategory:** Builder.io **License:** MIT **Source:** https://github.com/BuilderIO/skills/blob/main/skills/visual-recap/SKILL.md --- ## When it is useful Turn a PR, branch, commit, or git diff into an interactive visual recap with diagrams, file maps, API/schema summaries, annotated diffs, and review notes. ## The Catch Loud: Fusion Free forever $0 — 15 daily / 60 monthly Agent Credits; stops at cap; no rollover; proprietary SaaS (not OSS). Distinct from Expo leftovers / Firecrawl CLI. Skills MIT; product proprietary Fusion SaaS. --- # Builder WebMCP > Use a page MCP/WebMCP tools before UI automation **Subcategory:** Builder.io **License:** MIT **Source:** https://github.com/BuilderIO/skills/blob/main/skills/webmcp/SKILL.md --- ## When it is useful Open a user-provided URL in the host browser and use the page MCP or WebMCP tools before browser UI automation for app communication or edits. ## The Catch Loud: Fusion Free forever $0 — 15 daily / 60 monthly Agent Credits; stops at cap; no rollover; proprietary SaaS (not OSS). Distinct from Expo leftovers / Firecrawl CLI. Skills MIT; product proprietary Fusion SaaS. --- # Builder Agent Watchdog > Watch/audit/fix another agent session, PR, or transcript **Subcategory:** Builder.io **License:** MIT **Source:** https://github.com/BuilderIO/skills/blob/main/skills/agent-watchdog/SKILL.md --- ## When it is useful Watch, babysit, audit, review, compare, or fix another agent work from a session ID, transcript, chat link, PR, branch, log, or pasted run summary. ## The Catch Loud: Fusion Free forever $0 — 15 daily / 60 monthly Agent Credits; stops at cap; no rollover; proprietary SaaS (not OSS). Distinct from Expo leftovers / Firecrawl CLI. Skills MIT; product proprietary Fusion SaaS. HOLD plow-ahead, efficient-frontier, plan-arbiter, stay-within-limits, quick-recap, read-the-damn-docs, efficient-fable, rewind, adding-a-skill. --- # Flyte SDK Author > Scaffold Flyte 2 projects, tasks, workflows, launch plans **Subcategory:** Flyte **License:** Apache-2.0 **Source:** https://github.com/flyteorg/skills/blob/main/plugins/flyte/skills/flyte-sdk-author/SKILL.md --- ## When it is useful Create or scaffold Flyte workflows, tasks, or apps from scratch. ## The Catch Loud: Flyte OSS Apache-2.0 — local/devbox $0 forever; Flyte 1 self-host on your K8s $0. Official repo: Flyte 2 distributed OSS backend is coming soon — production Flyte 2 backend today is Union.ai (paid; do not invent free forever Cloud). Distinct from ZenML CA (MLOps pipelines vs K8s-native workflow orchestration). Skills Apache-2.0. --- # Flyte SDK Run > Run Flyte 2 workflows; logs, data, run lifecycle **Subcategory:** Flyte **License:** Apache-2.0 **Source:** https://github.com/flyteorg/skills/blob/main/plugins/flyte/skills/flyte-sdk-run/SKILL.md --- ## When it is useful Run Flyte 2 workflows, interact with runs and actions, retrieve logs and data, manage run lifecycle. ## The Catch Loud: Flyte OSS Apache-2.0 — local/devbox $0 forever; Flyte 1 self-host on your K8s $0. Official repo: Flyte 2 distributed OSS backend is coming soon — production Flyte 2 backend today is Union.ai (paid; do not invent free forever Cloud). Distinct from ZenML CA (MLOps pipelines vs K8s-native workflow orchestration). Skills Apache-2.0. --- # Flyte SDK Data > ETL, data quality, fanout/map task data patterns **Subcategory:** Flyte **License:** Apache-2.0 **Source:** https://github.com/flyteorg/skills/blob/main/plugins/flyte/skills/flyte-sdk-data/SKILL.md --- ## When it is useful Data engineering patterns: ETL pipelines, data processing, data quality checks, fanout/map tasks. ## The Catch Loud: Flyte OSS Apache-2.0 — local/devbox $0 forever; Flyte 1 self-host on your K8s $0. Official repo: Flyte 2 distributed OSS backend is coming soon — production Flyte 2 backend today is Union.ai (paid; do not invent free forever Cloud). Distinct from ZenML CA (MLOps pipelines vs K8s-native workflow orchestration). Skills Apache-2.0. --- # Flyte SDK ML > Training, HPO, experiment tracking on Flyte **Subcategory:** Flyte **License:** Apache-2.0 **Source:** https://github.com/flyteorg/skills/blob/main/plugins/flyte/skills/flyte-sdk-ml/SKILL.md --- ## When it is useful ML workload patterns: model training, hyperparameter optimization, experiment tracking. ## The Catch Loud: Flyte OSS Apache-2.0 — local/devbox $0 forever; Flyte 1 self-host on your K8s $0. Official repo: Flyte 2 distributed OSS backend is coming soon — production Flyte 2 backend today is Union.ai (paid; do not invent free forever Cloud). Distinct from ZenML CA (MLOps pipelines vs K8s-native workflow orchestration). Skills Apache-2.0. --- # Flyte SDK Ship > flyte.Image specs, Dockerfiles, dependency and tagging **Subcategory:** Flyte **License:** Apache-2.0 **Source:** https://github.com/flyteorg/skills/blob/main/plugins/flyte/skills/flyte-sdk-ship/SKILL.md --- ## When it is useful Generate flyte.Image specs, Dockerfiles, dependency management, image tagging, reproducible ship paths. ## The Catch Loud: Flyte OSS Apache-2.0 — local/devbox $0 forever; Flyte 1 self-host on your K8s $0. Official repo: Flyte 2 distributed OSS backend is coming soon — production Flyte 2 backend today is Union.ai (paid; do not invent free forever Cloud). Distinct from ZenML CA (MLOps pipelines vs K8s-native workflow orchestration). Skills Apache-2.0. --- # Running dbt Commands > Format and execute dbt CLI with the right executable **Subcategory:** dbt **License:** Apache-2.0 **Source:** https://github.com/dbt-labs/dbt-agent-skills/blob/main/skills/dbt/skills/running-dbt-commands/SKILL.md --- ## When it is useful Format and execute dbt CLI commands; select the correct dbt executable and flags. ## The Catch Loud: $0 forever path is dbt Core OSS Apache-2.0 (local/self-host). Platform Developer Free forever is real but capped (1 seat / 1 project / 3k successful models/mo); Starter $100/user/mo. Semantic Layer advanced is Cloud-paid — do not pitch Platform as the free hatch. Distinct from Flyte CD / ZenML CA. Skills Apache-2.0. --- # Adding dbt Unit Test > Unit test YAML mocking upstream model inputs **Subcategory:** dbt **License:** Apache-2.0 **Source:** https://github.com/dbt-labs/dbt-agent-skills/blob/main/skills/dbt/skills/adding-dbt-unit-test/SKILL.md --- ## When it is useful Create unit test YAML definitions that mock upstream model inputs and validate outputs. ## The Catch Loud: $0 forever path is dbt Core OSS Apache-2.0 (local/self-host). Platform Developer Free forever is real but capped (1 seat / 1 project / 3k successful models/mo); Starter $100/user/mo. Semantic Layer advanced is Cloud-paid — do not pitch Platform as the free hatch. Distinct from Flyte CD / ZenML CA. Skills Apache-2.0. --- # Building dbt Semantic Layer > Create or modify Semantic Layer components **Subcategory:** dbt **License:** Apache-2.0 **Source:** https://github.com/dbt-labs/dbt-agent-skills/blob/main/skills/dbt/skills/building-dbt-semantic-layer/SKILL.md --- ## When it is useful Create or modify dbt Semantic Layer components. Advanced Semantic Layer is Cloud-paid — Core/docs first. ## The Catch Loud: $0 forever path is dbt Core OSS Apache-2.0 (local/self-host). Platform Developer Free forever is real but capped (1 seat / 1 project / 3k successful models/mo); Starter $100/user/mo. Semantic Layer advanced is Cloud-paid — do not pitch Platform as the free hatch. Distinct from Flyte CD / ZenML CA. Skills Apache-2.0. --- # Configuring dbt MCP Server > MCP server JSON and auth for dbt **Subcategory:** dbt **License:** Apache-2.0 **Source:** https://github.com/dbt-labs/dbt-agent-skills/blob/main/skills/dbt/skills/configuring-dbt-mcp-server/SKILL.md --- ## When it is useful Generate MCP server configuration JSON and resolve authentication setup for dbt. ## The Catch Loud: $0 forever path is dbt Core OSS Apache-2.0 (local/self-host). Platform Developer Free forever is real but capped (1 seat / 1 project / 3k successful models/mo); Starter $100/user/mo. Semantic Layer advanced is Cloud-paid — do not pitch Platform as the free hatch. Distinct from Flyte CD / ZenML CA. Skills Apache-2.0. --- # Maintaining dbt Documentation > Audit coverage; draft missing model/column docs **Subcategory:** dbt **License:** Apache-2.0 **Source:** https://github.com/dbt-labs/dbt-agent-skills/blob/main/skills/dbt/skills/maintaining-dbt-documentation/SKILL.md --- ## When it is useful Audit dbt documentation coverage and draft missing model/column descriptions. ## The Catch Loud: $0 forever path is dbt Core OSS Apache-2.0 (local/self-host). Platform Developer Free forever is real but capped (1 seat / 1 project / 3k successful models/mo); Starter $100/user/mo. Semantic Layer advanced is Cloud-paid — do not pitch Platform as the free hatch. Distinct from Flyte CD / ZenML CA. Skills Apache-2.0. --- # Rivet Actors > Actors: the primitive for agent orchestration **Subcategory:** Rivet **License:** Apache-2.0 **Source:** https://github.com/rivet-dev/skills/blob/main/rivet-actors/SKILL.md --- ## When it is useful Use Rivet Actors as the primitive for agent orchestration. ## The Catch Loud: Free Cloud forever hard caps (100k Awake Actor Hours / 5GB / 5M writes / 200M reads / 100GB egress / $5 compute); OSS Apache self-host uncapped $0. Hobby $20+ / Team $200+. agentOS + Dynamic Apps = Beta. Skills LICENSE 404 — product rivet-dev/rivet Apache-2.0. Distinct from Flyte/Temporal/dbt. HOLD rivet-agentos/examples/browserbase browse. --- # Rivet agentOS > agentOS: a computer for every agent (Beta) **Subcategory:** Rivet **License:** Apache-2.0 **Source:** https://github.com/rivet-dev/skills/blob/main/rivet-agentos/SKILL.md --- ## When it is useful Use Rivet agentOS — a computer for every agent. Beta. ## The Catch Loud: Free Cloud forever hard caps (100k Awake Actor Hours / 5GB / 5M writes / 200M reads / 100GB egress / $5 compute); OSS Apache self-host uncapped $0. Hobby $20+ / Team $200+. agentOS + Dynamic Apps = Beta. Skills LICENSE 404 — product rivet-dev/rivet Apache-2.0. Distinct from Flyte/Temporal/dbt. HOLD rivet-agentos/examples/browserbase browse. --- # Rivet Dynamic Apps > Deploy an AI-generated backend per user (Beta) **Subcategory:** Rivet **License:** Apache-2.0 **Source:** https://github.com/rivet-dev/skills/blob/main/rivet-dynamic-apps/SKILL.md --- ## When it is useful Use Rivet Dynamic Apps to deploy an AI-generated backend for every user. Beta. ## The Catch Loud: Free Cloud forever hard caps (100k Awake Actor Hours / 5GB / 5M writes / 200M reads / 100GB egress / $5 compute); OSS Apache self-host uncapped $0. Hobby $20+ / Team $200+. agentOS + Dynamic Apps = Beta. Skills LICENSE 404 — product rivet-dev/rivet Apache-2.0. Distinct from Flyte/Temporal/dbt. HOLD rivet-agentos/examples/browserbase browse. --- # Rivet Workflows > Durable, replayable multi-step operations **Subcategory:** Rivet **License:** Apache-2.0 **Source:** https://github.com/rivet-dev/skills/blob/main/rivet-workflows/SKILL.md --- ## When it is useful Use Rivet Workflows for durable, replayable multi-step operations. ## The Catch Loud: Free Cloud forever hard caps (100k Awake Actor Hours / 5GB / 5M writes / 200M reads / 100GB egress / $5 compute); OSS Apache self-host uncapped $0. Hobby $20+ / Team $200+. agentOS + Dynamic Apps = Beta. Skills LICENSE 404 — product rivet-dev/rivet Apache-2.0. Distinct from Flyte/Temporal/dbt. HOLD rivet-agentos/examples/browserbase browse. --- # Polar Setup > Interactive onboarding wizard for Polar payments **Subcategory:** Polar **License:** Apache-2.0 **Source:** https://github.com/polarsource/skills/blob/main/skills/setup-polar/SKILL.md --- ## When it is useful Set up Polar from scratch: MCP, auth, first product, framework-specific checkout. ## The Catch Loud: Starter Free forever $0/mo platform fee; MoR take-rate 5% + 50c/txn (+1.5% intl) still applies — not payments free. Pro $20 / Growth $100 / Scale $400 lower take-rate. OSS Apache self-host path. Skills LICENSE 404 — product polarsource/polar Apache-2.0. Polar gem already live. Distinct from Stripe Billing DIY tax and closed MoRs. Frontmatter name setup-polar. --- # Polar Integration > Checkout, Customer Portal, and Webhooks via @polar-sh/sdk **Subcategory:** Polar **License:** Apache-2.0 **Source:** https://github.com/polarsource/skills/blob/main/skills/polar-integration/SKILL.md --- ## When it is useful Add Polar Checkout, Customer Portal, or Webhooks endpoints to a TypeScript/JavaScript app. ## The Catch Loud: Starter Free forever $0/mo platform fee; MoR take-rate 5% + 50c/txn (+1.5% intl) still applies — not payments free. Pro $20 / Growth $100 / Scale $400 lower take-rate. OSS Apache self-host path. Skills LICENSE 404 — product polarsource/polar Apache-2.0. Polar gem already live. Distinct from Stripe Billing DIY tax and closed MoRs. --- # Polar Migration > Migrate from Stripe Billing, Paddle, Lemon Squeezy, Gumroad **Subcategory:** Polar **License:** Apache-2.0 **Source:** https://github.com/polarsource/skills/blob/main/skills/polar-migration/SKILL.md --- ## When it is useful Plan or run a migration to Polar from another payment provider. ## The Catch Loud: Starter Free forever $0/mo platform fee; MoR take-rate 5% + 50c/txn (+1.5% intl) still applies — not payments free. Pro $20 / Growth $100 / Scale $400 lower take-rate. OSS Apache self-host path. Skills LICENSE 404 — product polarsource/polar Apache-2.0. Polar gem already live. Distinct from Stripe Billing DIY tax and closed MoRs. --- # Polar Testing > Sandbox, test cards, webhooks, CI for Polar **Subcategory:** Polar **License:** Apache-2.0 **Source:** https://github.com/polarsource/skills/blob/main/skills/polar-testing/SKILL.md --- ## When it is useful Test Polar integrations in sandbox: checkout, Stripe test cards, webhooks, CI. ## The Catch Loud: Starter Free forever $0/mo platform fee; MoR take-rate 5% + 50c/txn (+1.5% intl) still applies — not payments free. Pro $20 / Growth $100 / Scale $400 lower take-rate. OSS Apache self-host path. Skills LICENSE 404 — product polarsource/polar Apache-2.0. Polar gem already live. Distinct from Stripe Billing DIY tax and closed MoRs. --- # Gemini API Dev > Text, chat, multimodal, tools via Gemini API SDKs **Subcategory:** Gemini **License:** Apache-2.0 **Source:** https://github.com/google-gemini/gemini-skills/blob/main/skills/gemini-api-dev/SKILL.md --- ## When it is useful Write code that calls the Gemini API for text, chat, multimodal, image/video generation, streaming, function calling, or structured output. ## The Catch Loud: Gemini API Free Tier is real $0 (no CC) for eligible Flash models with hard RPM/RPD caps that change — do not invent numbers. Free-tier prompts may improve Google products. Pro/paid for higher quotas and production scale; Live/Omni video burn Free caps fast. Product proprietary Google SaaS — skills Apache-2.0, API is not OSS. Distinct from Firebase BO (ai-logic HOLD). --- # Gemini Live API Dev > Realtime bidirectional audio/video/text streaming **Subcategory:** Gemini **License:** Apache-2.0 **Source:** https://github.com/google-gemini/gemini-skills/blob/main/skills/gemini-live-api-dev/SKILL.md --- ## When it is useful Build realtime bidirectional streaming apps with the Gemini Live API (WebSocket audio/video/text, VAD, ephemeral tokens). ## The Catch Loud: Gemini API Free Tier is real $0 (no CC) for eligible Flash models with hard RPM/RPD caps that change — do not invent numbers. Free-tier prompts may improve Google products. Pro/paid for higher quotas and production scale; Live/Omni video burn Free caps fast. Product proprietary Google SaaS — skills Apache-2.0, API is not OSS. Distinct from Firebase BO (ai-logic HOLD). --- # Gemini Omni Flash API > Generative video edit/gen with Omni 1.1 Flash **Subcategory:** Gemini **License:** Apache-2.0 **Source:** https://github.com/google-gemini/gemini-skills/blob/main/skills/gemini-omni-flash-api/SKILL.md --- ## When it is useful Generative video editing, text-to-video, image-referenced video, frame transitions, and extensions with gemini-omni-1.1-flash. ## The Catch Loud: Gemini API Free Tier is real $0 (no CC) for eligible Flash models with hard RPM/RPD caps that change — do not invent numbers. Free-tier prompts may improve Google products. Pro/paid for higher quotas and production scale; Live/Omni video burn Free caps fast. Product proprietary Google SaaS — skills Apache-2.0, API is not OSS. Distinct from Firebase BO (ai-logic HOLD). --- # Statsig MCP > Inspect and manage gates, experiments, configs via Statsig MCP **Subcategory:** Statsig **License:** ISC **Source:** https://github.com/statsig-io/agent-skills/blob/main/skills/statsig/SKILL.md --- ## When it is useful Use the Statsig MCP to inspect and manage gates, experiments, dynamic configs, segments, metrics, audit logs, and results. ## The Catch Loud: Developer Free forever $0 no CC — 2M events/mo, unlimited seats, unlimited flag/config checks, 50k session replays/mo, 1-year analytics retention. Over Free → upgrade prompt to Pro ($150/mo + overages). Proprietary SaaS (not OSS); skills ISC. Distinct from LaunchDarkly CB and GrowthBook BC (OSS self-host). Frontmatter name statsig (parent). --- # Statsig Create Cloud Metric > Create Cloud metrics via Console API **Subcategory:** Statsig **License:** ISC **Source:** https://github.com/statsig-io/agent-skills/blob/main/skills/statsig-create-cloud-metric/SKILL.md --- ## When it is useful Create Statsig Cloud metrics via Console API POST /console/v1/metrics. ## The Catch Loud: Developer Free forever $0 no CC — 2M events/mo, unlimited seats, unlimited flag/config checks, 50k session replays/mo, 1-year analytics retention. Over Free → upgrade prompt to Pro ($150/mo + overages). Proprietary SaaS (not OSS); skills ISC. Distinct from LaunchDarkly CB and GrowthBook BC (OSS self-host). --- # Statsig Dashboard > Create or read dashboards via Console API **Subcategory:** Statsig **License:** ISC **Source:** https://github.com/statsig-io/agent-skills/blob/main/skills/statsig-dashboard/SKILL.md --- ## When it is useful Create or read Statsig dashboards through the Console API; build or update dashboard widgets. ## The Catch Loud: Developer Free forever $0 no CC — 2M events/mo, unlimited seats, unlimited flag/config checks, 50k session replays/mo, 1-year analytics retention. Over Free → upgrade prompt to Pro ($150/mo + overages). Proprietary SaaS (not OSS); skills ISC. Distinct from LaunchDarkly CB and GrowthBook BC (OSS self-host). --- # Ory Local Dev > Local Kratos/Keto/Hydra stack on :4000 without Network **Subcategory:** Ory **License:** Apache-2.0 **Source:** https://github.com/ory/claude-plugins/blob/master/plugins/ory-agent-plugin/skills/ory-local-dev/SKILL.md --- ## When it is useful Develop, prototype, or test login/registration/recovery/permission flows locally — run Ory offline, no Network project yet, spin up Kratos/Keto/Hydra. ## The Catch Loud: Network Developer Free $0 = PoC only (2 development envs, 0 prod/staging, 1 member). Production Network from ~$770/yr. Uncapped $0 = OSS self-host Apache-2.0 (Kratos/Keto/Hydra/Oathkeeper). Skills Apache-2.0 LICENSE 200. Aligns with Ory gem Evaluation framing. Distinct from Descope BD Cloud Free MAU. Ory gem already live. --- # Ory Auth Setup > Ory Network auth with Elements as default UI **Subcategory:** Ory **License:** Apache-2.0 **Source:** https://github.com/ory/claude-plugins/blob/master/plugins/ory-agent-plugin/skills/ory-auth-setup/SKILL.md --- ## When it is useful Add login, registration, recovery, email verification, settings, or sessions to a web app on Ory Network — Elements-first UI. ## The Catch Loud: Network Developer Free $0 = PoC only (2 development envs, 0 prod/staging, 1 member). Production Network from ~$770/yr. Uncapped $0 = OSS self-host Apache-2.0. Skills Apache-2.0 LICENSE 200. Aligns with Ory gem Catch. Distinct from Descope BD Cloud Free MAU. --- # Ory Login Flow > Elements pages for login, signup, recovery, settings **Subcategory:** Ory **License:** Apache-2.0 **Source:** https://github.com/ory/claude-plugins/blob/master/plugins/ory-agent-plugin/skills/ory-login-flow/SKILL.md --- ## When it is useful Build a login page, sign-up form, forgot-password flow, or account settings UI with Ory Elements self-service flows. ## The Catch Loud: Network Developer Free $0 = PoC only (2 development envs, 0 prod/staging, 1 member). Production Network from ~$770/yr. Uncapped $0 = OSS self-host Apache-2.0. Skills Apache-2.0 LICENSE 200. Distinct from Descope BD Cloud Free MAU. --- # Ory Social Login > OIDC providers (Google, GitHub, Apple, …) on Ory **Subcategory:** Ory **License:** Apache-2.0 **Source:** https://github.com/ory/claude-plugins/blob/master/plugins/ory-agent-plugin/skills/ory-social-login/SKILL.md --- ## When it is useful Configure Sign in with Google/GitHub/Apple/Microsoft/Discord/Slack/GitLab/Facebook on an Ory-powered app with Elements UI. ## The Catch Loud: Network Developer Free $0 = PoC only (2 development envs, 0 prod/staging, 1 member). Production Network from ~$770/yr. Uncapped $0 = OSS self-host Apache-2.0. Skills Apache-2.0 LICENSE 200. Distinct from Descope BD Cloud Free MAU. --- # Ory Permissions Onboarding > Agent-tool auth: observe → enforce, default-allow + deny **Subcategory:** Ory **License:** Apache-2.0 **Source:** https://github.com/ory/claude-plugins/blob/master/plugins/ory-agent-plugin/skills/ory-permissions-onboarding/SKILL.md --- ## When it is useful Onboard native agent-tool authorization — connect plugin, review allowed/blocked tools, write blocks in Console, promote observe to enforce. ## The Catch Loud: Network Developer Free $0 = PoC only (2 development envs, 0 prod/staging, 1 member). Production Network from ~$770/yr. Uncapped $0 = OSS self-host Apache-2.0 (incl. Keto). Skills Apache-2.0 LICENSE 200. Aligns with Ory gem Evaluation framing. Distinct from Descope BD Cloud Free MAU. --- # VoltAgent Create > CLI / manual bootstrap for VoltAgent agent projects **Subcategory:** VoltAgent **License:** MIT **Source:** https://github.com/VoltAgent/skills/blob/main/skills/create-voltagent/SKILL.md --- ## When it is useful Create a new AI agent project on the VoltAgent framework — CLI setup or manual bootstrapping. ## The Catch Loud: Core Framework Free forever MIT $0 (memory/RAG/guardrails/tools/MCP/voice/workflows). VoltOps Developer Free $0/mo (1 seat / 250 traces / 7-day retention); Core $50 / Pro $250. LLM BYOK. Skills MIT LICENSE 200. Distinct from Mastra (live gem), LangChain/LangGraph, Letta/CrewAI, Vercel AI SDK. CK VOID — do not restack MongoDB. Frontmatter name create-voltagent. --- # VoltAgent Best Practices > Agents vs workflows, layout, memory, servers, observability **Subcategory:** VoltAgent **License:** MIT **Source:** https://github.com/VoltAgent/skills/blob/main/skills/voltagent-best-practices/SKILL.md --- ## When it is useful Apply VoltAgent architectural patterns — agents vs workflows, project layout, memory, servers, observability. ## The Catch Loud: Core Framework Free forever MIT $0 (memory/RAG/guardrails/tools/MCP/voice/workflows). VoltOps Developer Free $0/mo (1 seat / 250 traces / 7-day retention); Core $50 / Pro $250. LLM BYOK. Skills MIT LICENSE 200. Distinct from Mastra (live gem), LangChain/LangGraph, Letta/CrewAI, Vercel AI SDK. CK VOID — do not restack MongoDB. Frontmatter name voltagent-best-practices. --- # VoltAgent Core Reference > VoltAgent class: constructor, lifecycle, runtime **Subcategory:** VoltAgent **License:** MIT **Source:** https://github.com/VoltAgent/skills/blob/main/skills/voltagent-core-reference/SKILL.md --- ## When it is useful Look up VoltAgent class constructor options, lifecycle methods, and runtime behavior. ## The Catch Loud: Core Framework Free forever MIT $0 (memory/RAG/guardrails/tools/MCP/voice/workflows). VoltOps Developer Free $0/mo (1 seat / 250 traces / 7-day retention); Core $50 / Pro $250. LLM BYOK. Skills MIT LICENSE 200. Distinct from Mastra (live gem), LangChain/LangGraph, Letta/CrewAI, Vercel AI SDK. CK VOID — do not restack MongoDB. Frontmatter name voltagent-core-reference. --- # VoltAgent Docs Bundle > Version-matched docs from @voltagent/core/docs **Subcategory:** VoltAgent **License:** MIT **Source:** https://github.com/VoltAgent/skills/blob/main/skills/voltagent-docs-bundle/SKILL.md --- ## When it is useful Look up VoltAgent API signatures, guides, and examples from node_modules/@voltagent/core/docs (version-matched). ## The Catch Loud: Core Framework Free forever MIT $0 (memory/RAG/guardrails/tools/MCP/voice/workflows). VoltOps Developer Free $0/mo (1 seat / 250 traces / 7-day retention); Core $50 / Pro $250. LLM BYOK. Skills MIT LICENSE 200. Distinct from Mastra (live gem), LangChain/LangGraph, Letta/CrewAI, Vercel AI SDK. CK VOID — do not restack MongoDB. Frontmatter name voltagent-docs-bundle. --- # CircleCI Builds > Diagnose failing jobs and flaky pipelines **Subcategory:** CircleCI **License:** MIT **Source:** https://github.com/CircleCI-Public/skills/blob/main/plugins/circleci/skills/builds/SKILL.md --- ## When it is useful Investigate failed CircleCI jobs, triage flaky pipelines, find root causes from logs, and apply minimal config/test/build fixes. ## The Catch Loud: Free plan $0/mo no CC — 30k credits/mo (~up to 6k small Docker mins), 5 users; Free credits expire monthly (no rollover). OSS orgs on Free up to 400k credits/mo Linux/Arm/Docker. Performance from ~$15/mo. Proprietary SaaS CI — not self-host OSS core. Skills root LICENSE 404; plugin.json MIT + circleci-cli MIT. Distinct from Buildkite BB (agent pipelines) and Flux BE (GitOps). HOLD chunk + smarter-testing. Frontmatter name circleci-builds. --- # CircleCI CLI > Auth, validate, watch runs, logs, rerun/cancel **Subcategory:** CircleCI **License:** MIT **Source:** https://github.com/CircleCI-Public/skills/blob/main/plugins/circleci/skills/cli/SKILL.md --- ## When it is useful Day-to-day CircleCI from the terminal — authenticate, validate config, watch a run, read logs/failed tests, download artifacts, rerun or cancel. ## The Catch Loud: Free plan $0/mo no CC — 30k credits/mo (~up to 6k small Docker mins), 5 users; Free credits expire monthly (no rollover). OSS orgs on Free up to 400k credits/mo Linux/Arm/Docker. Performance from ~$15/mo. Proprietary SaaS CI — not self-host OSS core. Skills root LICENSE 404; plugin.json MIT + circleci-cli MIT. Distinct from Buildkite BB (agent pipelines) and Flux BE (GitOps). HOLD chunk + smarter-testing. Frontmatter name circleci-cli. --- # CircleCI Config > Optimize .circleci/config.yml cache and parallelism **Subcategory:** CircleCI **License:** MIT **Source:** https://github.com/CircleCI-Public/skills/blob/main/plugins/circleci/skills/config/SKILL.md --- ## When it is useful Improve `.circleci/config.yml` for speed and reliability — caching, workspaces, parallelism, or config-caused flake. ## The Catch Loud: Free plan $0/mo no CC — 30k credits/mo (~up to 6k small Docker mins), 5 users; Free credits expire monthly (no rollover). OSS orgs on Free up to 400k credits/mo Linux/Arm/Docker. Performance from ~$15/mo. Proprietary SaaS CI — not self-host OSS core. Skills root LICENSE 404; plugin.json MIT + circleci-cli MIT. Distinct from Buildkite BB (agent pipelines) and Flux BE (GitOps). HOLD chunk + smarter-testing. Frontmatter name circleci-config. --- # CircleCI Onboard > Account → org → GitHub App → project → pipeline **Subcategory:** CircleCI **License:** MIT **Source:** https://github.com/CircleCI-Public/skills/blob/main/plugins/circleci/skills/onboard/SKILL.md --- ## When it is useful First-time CircleCI setup — account, org, GitHub App, project, pipeline definition, iterate config until it passes. ## The Catch Loud: Free plan $0/mo no CC — 30k credits/mo (~up to 6k small Docker mins), 5 users; Free credits expire monthly (no rollover). OSS orgs on Free up to 400k credits/mo Linux/Arm/Docker. Performance from ~$15/mo. Proprietary SaaS CI — not self-host OSS core. Skills root LICENSE 404; plugin.json MIT + circleci-cli MIT. Distinct from Buildkite BB (agent pipelines) and Flux BE (GitOps). HOLD chunk + smarter-testing. Frontmatter name onboarding. --- # CircleCI Testsuite > .circleci/test-suites.yml + testsuite doctor **Subcategory:** CircleCI **License:** MIT **Source:** https://github.com/CircleCI-Public/skills/blob/main/plugins/circleci/skills/testsuite/SKILL.md --- ## When it is useful Onboard or migrate tests onto CircleCI testsuite (`.circleci/test-suites.yml` + `circleci testsuite run`) via testsuite doctor — not paid smarter-testing features. ## The Catch Loud: Free plan $0/mo no CC — 30k credits/mo (~up to 6k small Docker mins), 5 users; Free credits expire monthly (no rollover). OSS orgs on Free up to 400k credits/mo Linux/Arm/Docker. Performance from ~$15/mo. Proprietary SaaS CI — not self-host OSS core. Skills root LICENSE 404; plugin.json MIT + circleci-cli MIT. Distinct from Buildkite BB (agent pipelines) and Flux BE (GitOps). HOLD chunk + smarter-testing. Frontmatter name circleci-testsuite. --- # Xata Using API > HTTP API for orgs, projects, branches, auth **Subcategory:** Xata **License:** Apache-2.0 **Source:** https://github.com/xataio/skills/blob/main/skills/using-xata-api/SKILL.md --- ## When it is useful Call the Xata HTTP API (curl/fetch/SDK) to manage orgs, projects, branches, auth, connection strings, pagination, and errors. ## The Catch Loud: $0 forever path is Open Source self-host (branching/scale-to-zero) — verified xata.io/pricing. Xata Cloud is 14-day free trial then PAYG from ~$0.012/hr + $0.28/GB/mo storage — not a forever free Cloud hatch. Skills Apache-2.0 LICENSE 200. Distinct from Neon/Supabase/Postgres gems and Redis BH. Frontmatter name using-xata-api. --- # Xata Using CLI > xata CLI, pgroll, scratch branches, metrics **Subcategory:** Xata **License:** Apache-2.0 **Source:** https://github.com/xataio/skills/blob/main/skills/using-xata-cli/SKILL.md --- ## When it is useful Manage Xata via the xata CLI — projects, branches, pgroll migrations, scratch branches, metrics, logs, connection strings. ## The Catch Loud: $0 forever path is Open Source self-host (branching/scale-to-zero) — verified xata.io/pricing. Xata Cloud is 14-day free trial then PAYG from ~$0.012/hr + $0.28/GB/mo storage — not a forever free Cloud hatch. Skills Apache-2.0 LICENSE 200. Distinct from Neon/Supabase/Postgres gems and Redis BH. Frontmatter name using-xata-cli. --- # Xata Managing PostgreSQL > PG ops and performance in a Xata branch **Subcategory:** Xata **License:** Apache-2.0 **Source:** https://github.com/xataio/skills/blob/main/skills/managing-postgresql/SKILL.md --- ## When it is useful Diagnose PostgreSQL issues in a Xata branch context — slow queries, locks, vacuum, indexes, connections, replication, backups. ## The Catch Loud: $0 forever path is Open Source self-host (branching/scale-to-zero) — verified xata.io/pricing. Xata Cloud is 14-day free trial then PAYG from ~$0.012/hr + $0.28/GB/mo storage — not a forever free Cloud hatch. Skills Apache-2.0 LICENSE 200. Distinct from Neon/Supabase/Postgres gems and Redis BH. Frontmatter name managing-postgresql. --- # About Stack Gems Stack Gems is an agent-curated catalog of developer tools. Every tool listed is real, every URL is verified, and every review represents a genuine assessment of the tool's strengths and tradeoffs. Human editors maintain veto power over all content. The catalog expands daily as new tools emerge. Submit tools for review at https://stackgems.com/submit