Observability tools

• LangfuseOpen-source LLM observability you can self-host

  Free / OSS · Open-Core · 36k GitHub stars

  VerdictThe LLM observability platform for teams who want to own their data.

  CatchLicense: the core is MIT, but anything under the ee/ directories is under a separate enterprise license — check which features you rely on before calling your deployment fully MIT. Self-hosting means self-operating—you are running Postgres, ClickHouse, Redis, and the app. Observability-first means evals are layered on, not central.

• GrafanaDashboards for everything observable

  Free / OSS · AGPL-3.0 · 77.4k GitHub stars

  VerdictThe universal observability dashboard.

  CatchDashboard sprawl becomes a maintenance burden. Learning PromQL or LogQL takes time. Self-hosting requires infrastructure.

• PrometheusTime-series metrics for cloud-native

  Free / OSS · Apache-2.0 · 66.6k GitHub stars

  VerdictThe foundation of cloud-native metrics.

  CatchSingle-node storage limits scale. PromQL has a learning curve. High cardinality labels cause problems.

• OpenTelemetryVendor-neutral observability framework

  Free / OSS · OSS/MIT

  VerdictThe standard for future-proof instrumentation.

  CatchThe specification is complex. Implementation maturity varies by language. Configuration requires understanding the model.

• UnleashAGPLv3 feature flags; OSS caps 1 project / 2 environments

  Free / OSS · AGPL-3.0 · 13.9k GitHub stars

  VerdictAGPLv3 flags. Self-host with OSS caps; Cloud trial is not free forever.

  CatchOSS caps Projects 1 / Environments 2. Cloud = Enterprise trial FAIL — pitch self-host only. Vs PostHog dedicated-flags-inside-analytics. Distinct from PostHog.

• FlagsmithBSD-3-Clause flags; Cloud Free 50k req/mo, self-host uncapped

  Free / OSS · Freemium · 6.6k GitHub stars

  VerdictBSD-3 flags. Cloud Free is real but capped; self-host for scale.

  CatchCloud 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.

• DeepEvalApache-2.0 pytest-style LLM/agent evals; Confident AI is the paid cloud

  Free / OSS · Apache-2.0 · 18.9k GitHub stars

  VerdictApache-2.0 LLM unit tests. Free locally; Confident AI starts free then $200.

  CatchOSS 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.

• Uptime KumaMIT self-host uptime monitoring and status pages

  Free / OSS · OSS/MIT · 92.6k GitHub stars

  VerdictMIT self-host uptime. Free forever; you own the box on 3001.

  CatchSelf-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.

• VictoriaMetricsApache-2.0 Prom-compatible TSDB storage swap, not full APM UI

  Free / OSS · Apache-2.0 · 17.9k GitHub stars

  VerdictApache-2.0 Prom-compatible TSDB. CE free; downsampling/anomaly/multi-tenant Enterprise.

  CatchStorage/metrics engine - not Grafana/SigNoz APM UI. Downsampling, anomaly detection (vmanomaly), and advanced multi-tenant stats/rate-limit bits are Enterprise. Distinct from Prometheus encyclopedia and SigNoz all-in-one.

• VectorMPL-2.0 logs/metrics/traces data plane vs OTel Collector/Alloy

  Free / OSS · MPL-2.0 · 22.7k GitHub stars

  VerdictMPL-2.0 obs data plane. Free to run; watch MPL + Datadog stewardship.

  CatchMPL-2.0 (file-level copyleft, not MIT/Apache). Datadog-maintained after Timber acquisition - product direction follows Datadog. Distinct from OpenTelemetry Collector and Grafana Alloy.

• UmamiMIT cookieless analytics; Cloud Hobby is a real free tier

  Free / OSS · OSS/MIT · 39.5k GitHub stars

  VerdictMIT analytics. Free forever self-host; Cloud Hobby if you want $0 managed for a single site.

  CatchNot PostHog/GA depth — no session replay/heatmaps on Hobby/self-host the way Cloud Business sells them. Self-host means you own updates, backups, and a growing DB. Vs Plausible (paid SaaS-first) and PostHog.

• ctrlb-decomposeMIT Drain3-style log pattern collapse; logs stay local (Rust/WASM)

  Free / OSS · OSS/MIT · 305 GitHub stars

  VerdictMIT local log→pattern tool. Free forever; use when logs must stay on-box before an LLM sees them.

  CatchVs Vector (ships/transforms telemetry), SigNoz/Grafana (UIs): ctrlb-decompose is the *local pattern collapser*, not a collector and not a dashboard. You still decide what runs after the patterns (agent prompt, ticket, or your own store). Early open release — treat CLI flags and WASM embed path as moving.

• k6AGPL-3.0 scriptable load testing (JS) from Grafana

  Free / OSS · AGPL-3.0 · 32k GitHub stars

  VerdictAGPL-3.0 scriptable load. Free unmodified CLI/CI; Cloud k6 is the paid seat.

  CatchAGPL-3.0 network/copyleft — modifications you serve or distribute can trigger source obligations; do not treat k6 as MIT. Grafana Cloud k6 is a separate paid hosted product (execution, analysis, collaboration). Vs oha (MIT TUI hammer): k6 is the *scripted/scenario* suite, not a live ratatui chart. Point it at staging you own.

• sternApache-2.0 multi-pod Kubernetes log tail

  Free / OSS · Apache-2.0 · 4.9k GitHub stars

  VerdictApache-2.0 K8s log tail. Free forever; follow the fleet, not one pod at a time.

  CatchVs k9s (full Kubernetes *TUI*): stern is the *multi-pod log tail CLI*, not a cluster navigator, not Helm, not a dashboard. Needs kubeconfig and RBAC to read logs. Not Loki/Vector (log pipelines).

• OpenAI Agents SDKMIT multi-agent workflows — tools, handoffs, tracing, MCP

  Free / OSS · OSS/MIT · 30.1k GitHub stars

  VerdictMIT OpenAI agent workflow SDK. Build multi-agent loops; API spend is separate.

  CatchVs openai-codex-cli (terminal coding agent), Vercel AI SDK (app AI primitives), and Microsoft Agent Framework: this is the workflow SDK, not the Codex CLI seat. OpenAI API keys and hosted tools meter; sandbox/realtime extras add surface area. Do not confuse with ChatGPT product seats.

• DozzleMIT realtime Docker/Swarm/K8s log UI

  Free / OSS · OSS/MIT · 14.7k GitHub stars

  VerdictMIT live container log UI. Lightweight tail vs full observability stacks.

  CatchVs Grafana/Loki, ELK, Datadog (full observability): Dozzle is lightweight realtime log UI, not metrics/traces/storage/retention. If the container is gone, so is the history Dozzle never kept. Pair with a real log backend when you need audit trails.

• OpenStatusAGPL-3.0 status/uptime as code — Terraform/CLI/MCP

  Free / OSS · Freemium · 9.2k GitHub stars

  VerdictAGPL status/uptime as code. Free Hobby is tiny; self-host or pay when monitors multiply.

  CatchAGPL-3.0 — network copyleft on modifications you ship as a service. free cloud: one monitor / one page / 10m checks — fine for a side project, not a fleet. Vs Better Stack / Checkly-class SaaS, OpenStatus wins on OSS+IaC; you trade managed depth for license+self-host ops.

• SLayerMIT embeddable semantic layer for agents — MCP, SQL expressions, dbt/Cube import

  Free / OSS · OSS/MIT · 303 GitHub stars

  VerdictOSS semantic layer for agentic analytics. Expressions + MCP over your warehouse; not another vector RAG pack.

  CatchVs warehouse/BI tools and raw SQL MCP: SLayer is the metric/expression layer agents query, not a database or a full BI product. You still own warehouse credentials, row-level security design, and whether definitions stay truthful. Young Motley surface — verify connectors for your engine before betting a production agent on it.

• AURAApache-2.0 MCP-ready SRE agent platform for production incidents

  Free / OSS · Apache-2.0 · 399 GitHub stars

  VerdictApache-2.0 SRE agent platform. Use for production incident investigation with MCP and your own models; wire real telemetry first.

  CatchVs generic coding agents: AURA is an SRE/incident harness, not a repo coding ADE. Deepest value shows when wired to real observability backends (Mezmo and others); empty integrations will not invent incidents. Hosted Mezmo sessions are optional — self-host is the OSS path.

• CiliumeBPF networking, security, and observability for Kubernetes

  Free / OSS · Apache-2.0 · 25.7k GitHub stars

  VerdictApache-2.0 eBPF Kubernetes networking. Default when identity-aware policy and Hubble flows beat plain CNI.

  CatchVs flannel/Calico defaults: Cilium is the deeper eBPF control plane — more power, more kernel/datapath expertise. Isovalent/Cisco commercial layers exist around the same codebase; self-host OSS carefully for upgrades and Hubble ops.

• ContextForgeApache MCP+A2A+REST/gRPC→MCP gateway with governance + OTel

  Free / OSS · Apache-2.0 · 4.7k GitHub stars

  VerdictApache MCP/A2A gateway when you need federation + governance + OTel on your own infra and will pay the ops cost; skip if you only needed a thin remote MCP shim.

  CatchVs SuperGateway and mcp-remote: ContextForge is a heavier self-host governance/observability gateway, not a thin MCP bridge. Vs Smithery: this is a self-host gate you operate, not a hosted MCP marketplace. Open-issue count is high for the maturity pitch; local/JWT setups need real secrets — do not ship example tokens.

• tRPC-Agent-GoApache-2.0 Go production agent framework — GraphAgent, MCP, A2A

  Free / OSS · Apache-2.0 · 1.9k GitHub stars

  VerdictApache-2.0 Go agent framework when GraphAgent + MCP/A2A belong inside Go services — skip if your team is Python-first and does not want another runtime.

  CatchGo-native stack with tRPC-ecosystem gravity — not a drop-in for Python LangGraph/Crew shops. Not a hosted product: you build and run the service. Pick it when Go is already the runtime, not as a reason to leave a working Python agent stack.

• Arize PhoenixELv2 self-host AI observability & eval — traces, datasets, playground

  Free / OSS · Open-Core · 11.9k GitHub stars

  VerdictELv2 self-host AI obs/eval when Phoenix’s tracing+evals fit and you accept source-available — not Langfuse MIT, not Braintrust CI-eval.

  CatchNOT OSI open source — ELv2 forbids offering Phoenix as a competing hosted SaaS. Vs Langfuse (MIT full-stack self-host) and Braintrust (eval-first freemium): Phoenix is Arize’s ELv2 observability/eval kit with strong notebook/embedding UX, not MIT and not Braintrust’s CI-eval product. Arize cloud is the upsell when you want managed multi-tenant.

• Mobile Build MCPMIT Sentry MCP+CLI for iOS/macOS agent builds — npm mobilebuildmcp

  Free / OSS · OSS/MIT · 6.5k GitHub stars

  VerdictMIT Apple build MCP when agents must compile iOS/macOS — not mobile-mcp a11y driving and not browser MCPs.

  CatchVs mobile-mcp / chrome-devtools-mcp / playwright-mcp: mobile-mcp is Apache native iOS+Android a11y-tree control (sim/emu/real); DevTools/Playwright are browser — Mobile Build MCP’s catch is build/compile tooling for Apple projects, not UI automation. macOS + Xcode required; Android is out of scope; ★6448 is healthy but still early — pin a release before fleet install.

• HeadroomApache context-compression layer — shrink tool dumps before the model

  Free / OSS · Apache-2.0 · 75.9k GitHub stars

  VerdictApache context compression when tool dumps eat the window — not LiteLLM proxy, not Langfuse, not Mem0; treat the star spike with caution; verify FATAL lines survive.

  Catch★74201 in about 38 weeks (~1953 a week): judge it on the compression fidelity and failure modes, not the star count. Vs litellm / langfuse / mem0: those are proxy/obs/memory lanes — Headroom’s catch is lossy/lossless-ish context compression of tool/RAG payloads. Always validate that error lines and stack traces you care about survive; pin `headroom-ai` package names (not unrelated `headroom` on PyPI).

• AgentgatewayApache AI-native proxy — MCP & A2A security, obs, governance

  Free / OSS · Apache-2.0 · 5.4k GitHub stars

  VerdictApache MCP/A2A proxy when agents need governed tool and agent traffic — not LiteLLM/Weave/CCR model routers and not Nasiko harness control.

  CatchVs litellm / weave-router / claude-code-router / nasiko: LiteLLM is a multi-provider LLM proxy; Weave/CCR are coding-agent model routers; Nasiko is harness discover/route/trace — Agentgateway’s catch is MCP/A2A agentic connectivity proxy (security/obs/governance), not model routing alone. Ops weight of a full proxy; pin a release and map which MCP/A2A clients you actually terminate.

• BeszelMIT lightweight server + Docker monitoring — PocketBase hub, tiny agents, alerts

  Free / OSS · OSS/MIT · 26.3k GitHub stars

  VerdictMIT self-hosted server and container monitoring when Prometheus + Grafana is too much — free, quick to run, but a fixed metric set on a single-hub design.

  CatchFixed set of host and container metrics, not a general metrics platform: no PromQL, no custom app metrics, no logs or traces. The hub is one PocketBase (SQLite) app, so think small fleets, not hundreds of nodes with long high-resolution retention. The agent needs host access (and the Docker/Podman socket for container stats), so scope it like any privileged daemon. Vs Uptime Kuma (endpoint uptime), Dozzle (container logs) and Prometheus + Grafana (full metrics stack): Beszel is the resource-history-and-alerts middle seat.

• Grafana MCP serverApache-2.0 MCP server from Grafana Labs for dashboards, datasources, alerts and incidents

  Free / OSS · Apache-2.0 · 3.6k GitHub stars

  VerdictThe official way to put Grafana in an agent’s hands — enable only the tool groups you need and keep dashboard edits small.

  CatchNeeds Grafana 9.0 or newer. Several tool groups — panel queries, SQL datasources, InfluxDB, CloudWatch and more — stay off until you turn them on with `--enabled-tools`, so check the list before assuming the agent can see a datasource. Updating a dashboard means sending its full JSON, which eats a lot of agent context on big dashboards. Vs grafana (the dashboards themselves) and opensre (an agent that runs incident investigations end to end): this is the plain bridge that gives any agent read and write access to Grafana.

• HigressApache-2.0 AI and API gateway on Envoy and Istio, with Wasm plugins, model proxying and MCP hosting

  Free / OSS · Apache-2.0 · 9.6k GitHub stars

  VerdictA production-grade AI gateway if you can run Envoy and Istio — for a quick model proxy, start with something smaller.

  CatchIt is a full Envoy/Istio gateway, much heavier to run than a single-binary model proxy. Default images pull from an Alibaba Cloud registry in Hangzhou, so outside China switch to the US or Southeast Asia mirror if pulls time out. Much of the documentation and the plugin catalog live on higress.cn. Vs agentgateway (a lighter, purpose-built gateway for agent and MCP traffic) and litellm (a Python model proxy you can start in minutes): Higress fits when you want AI traffic inside a production API gateway you already run at scale.

• MCP Toolbox for DatabasesApache-2.0 Google MCP server for databases

  Free / OSS · Apache-2.0 · 16.7k GitHub stars

  VerdictThe Google-backed way to give agents database access through queries you control — budget time for writing tools.yaml.

  CatchYou write and maintain your SQL tools yourself in tools.yaml — the server does not invent safe queries for you. The prebuilt tool sets mostly target Google databases (AlloyDB, Cloud SQL, BigQuery, Spanner and friends); other engines get the generic tools. Renamed from genai-toolbox, so older guides use the old name.

More with this tag (46)

• Supabase MCP Server — Apache-2.0 official MCP server for Supabase
• OpenTelemetry Collector Contrib — Apache-2.0 extra OTel components
• Jaeger — Apache-2.0 CNCF distributed tracing, OTLP-native
• node_exporter — Apache-2.0 official Prometheus host-metrics exporter
• Apache Maka — Apache-2.0 local-first agent workspace that logs every step
• Nginx UI — AGPL-3.0 web UI for managing Nginx configs, certs and logs
• Opik — Apache-2.0 LLM and agent tracing, evaluation and monitoring by Comet
• Gatus — Apache-2.0 health checks and status page with alerts
• CrowdSec — MIT open-source IDS/IPS and WAF with a community blocklist
• Termix — Apache-2.0 self-hosted SSH and server manager
• Kener — MIT status page system built with SvelteKit
• Grafana Pyroscope — AGPL-3.0 continuous profiling platform
• HyperDX — MIT logs, traces and session replay search on ClickHouse
• Coroot — Apache-2.0 eBPF observability with service maps
• OpenLLMetry — Apache-2.0 OpenTelemetry instrumentation for LLM apps
• GreptimeDB — Apache-2.0 metrics, logs and traces database
• Pixie — Apache-2.0 eBPF in-cluster observability for Kubernetes
• Grafana Alloy — Apache-2.0 OpenTelemetry Collector distribution
• Grafana Mimir — AGPL-3.0 scalable long-term Prometheus storage
• Parca — Apache-2.0 eBPF continuous profiling down to the line number
• LangWatch — Apache-2.0 LLM tracing, agent tests, evals, gateway
• Uptrace — AGPL-3.0 OpenTelemetry APM on ClickHouse and Postgres
• Odigos — Apache-2.0 distributed tracing without code changes
• HolmesGPT — Apache-2.0 CNCF AI agent for production incidents
• Laminar — Apache-2.0 tracing, evals and signals for AI agents
• Perses — Apache-2.0 CNCF dashboards with an open dashboard spec
• Sloth — Apache-2.0 SLO generator for Prometheus rules and alerts
• Kubetail — Apache-2.0 real-time log dashboard for Kubernetes workloads
• VictoriaLogs — Apache-2.0 zero-config schema-free log database
• Parseable — AGPL-3.0 observability data lake in a single binary
• OpenLIT — Apache-2.0 OTel tracing, evals and guardrails for agents
• Nightingale — Apache-2.0 alerting hub for your existing data sources
• Alertmanager — Apache-2.0 Prometheus alert routing and silencing
• Kratos — MIT Go framework for cloud-native microservices
• Versus Incident — MIT self-hosted incident platform with AI SRE agent
• Apache HertzBeat — Apache-2.0 agentless monitoring, alerting and status pages in one
• Ragas — Apache-2.0 RAG/agent eval metrics — synthetic data + CI scorers
• Sentry — Error tracking that actually helps
• Axiom — Log management without limits
• Better Stack — Uptime monitoring and incident management
• Highlight.io — Open-source session replay and monitoring
• PostHog — Product analytics, flags, replay, and errors in one MIT-core stack
• GrowthBook — MIT-core flags+experiments; Cloud Starter Free, self-host unlimited users
• SigNoz — MIT-core OTel APM (logs/metrics/traces); Cloud from $49/mo
• Datadog — Unified observability platform
• LangSmith — LangChain freemium obs/eval/deploy — free Developer 5k traces/mo