For FastAPI developers
Same decorators. Agent superpowers.
FastAgentic is a FastAPI extension. If you can write @app.post("/"), you can write @agent_endpoint("/"). Everything you know about dependency injection, Pydantic models, middleware, and testing still applies.
What you get beyond FastAPI
- Automatic MCP + A2A surfaces — the same function becomes an MCP tool and A2A skill. No duplicate schemas.
- Streaming without boilerplate — SSE and WebSocket responses from an async generator, with backpressure handled.
- run_opaque() — wrap existing agent code and get crash-safe resumption without refactoring.
- Cost tracking middleware — token counts per request, attributed to user/tenant.
- Policy engine — declarative rate limits and budget caps that compose like FastAPI dependencies.
What stays the same
- Pydantic 2 request/response models.
Depends(),BackgroundTasks,Request,Response— unchanged.- OpenAPI/Swagger generation.
- Uvicorn + Gunicorn deployment.
- pytest + httpx testing.
Migration path from raw FastAPI
- Install:
pip install fastagentic[pydanticai] - Replace your
FastAPI()app withApp(). - Swap
@app.post→@agent_endpointfor agent-bearing routes. - Point your storage adapter at Redis or Postgres for durability.
- Keep everything else.
Migrating a large FastAPI codebase to agents?
Neul Labs — the team behind FastAgentic — takes on a limited number of consulting engagements each quarter. We help teams ship agents to production, fix broken LangGraph pipelines, and design governance for multi-tenant LLM platforms.