Orchestration for Python

Haystack strength is also its constraint. The framework assumes pipeline-based architectures; more exotic agent patterns require workarounds.

Vs OpenRouter/provider SDKs: self-host gateway you operate — not a hosted model marketplace. Enterprise/ tree and SSO/audit/RBAC are commercial; core proxy stays MIT. You still pay model providers.

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.

Vs LangGraph (graphs), CrewAI (roles), Mastra (TypeScript): Agno is Python performance + AgentOS productization, not a coding CLI like Claude Code/OpenCode. Phidata docs/imports are dead—use Agno. Control Plane and hosted conveniences are optional; day-one is OSS SDK+runtime you operate.

Vs microsoft/autogen (maintenance; MS points new work to Microsoft Agent Framework) and LangGraph/CrewAI: AG2 is the community AutoGen fork, not Microsoft’s enterprise successor. AG2 v1 is not a drop-in from Classic — pin ag2-classic / import autogen if you need the old surface. One primary entry; do not twin with dead AutoGen marketing.

Vs github-mcp (one official GitHub surface) and raw MCP servers: Composio is a multi-app auth/tool platform — you depend on their connector quality and hosted auth. Free tier exists; production is seat/usage freemium. Not a coding agent and not a browser fleet.

Vs Gemini CLI / Antigravity CLI (terminal coding agents), CrewAI/LangGraph/Agno (orchestration frameworks), and Daytona/E2B (sandboxes): ADK is the Google build/eval/deploy kit, not a coding CLI and not a sandbox fleet. Vertex/GCP gravity is real for enterprise deploy; OSS still runs without a Google Cloud bill if you stay local/BYO.

Vs LangChain/LangGraph (orchestration graphs) and Promptfoo (YAML eval/red-team): DSPy is program-and-optimize LMs, not an MCP-first runtime and not a hosted eval SaaS. Optimizer runs cost tokens/compute; you still bring models and data. Not a drop-in coding agent.

Vs CrewAI / AG2 (product multi-agent orchestration) and OpenHands / SWE-agent (coding agents): CAMEL is multi-agent research/framework breadth, not a turnkey SWE seat. Eigent is a separate commercial cousin — do not treat CAMEL OSS as Eigent cloud. You own model spend and orchestration choices.

Vs OpenHands, LangGraph, and Mastra: DeerFlow is a full SuperAgent product harness with sandboxes/memory/gateway, not an open coding agent alone, a graph orchestration library, or a TypeScript workflow framework. BYO model API keys; Docker sizing is resource-heavy (multi-vCPU/GB guidance); star velocity is extreme (~83k).

Vs CrewAI and AG2: this is MCP-native multi-agent + shared graph for coding workflows, not a general role-orchestration framework alone. LICENSE/README are AGPL-3.0 while pyproject (and node metadata) still claim MIT — metadata conflict; treat AGPL as authoritative until packaging is fixed. Last push 2026-03-28; typical path expects an OpenAI key.

★8499 in about 41 weeks (~207 a week): judge it on the sandbox shape, not the star count. Vs E2B / AIO Sandbox / CubeSandbox: those are hosted Firecracker seats, broader sandbox/browser/MCP runtimes, or self-hosted RustVMM/KVM density — Monty’s catch is in-process Rust Python sandbox from Pydantic, not a sandbox SaaS seat. Do not confuse OSS Monty packages with Full Monty commercial; pin the MIT crates/npm/PyPI names.

★155k: a large, mature project, but stars still are not a quality signal. Vs Dify / n8n / mcp-agent: Dify is OSS+cloud LLM studio; n8n is fair-code general automation; mcp-agent is an agent framework — Langflow’s catch is the agent/RAG visual builder with first-class MCP tool export. Self-host ops and dependency weight are real; pin OSS deploy docs before promising zero-ops.