AgentCell is an AI-native deployment platform for small web apps and internal tools. This sample is FastAPI + uvicorn — a small in-memory JSON API plus one HTML page, the ASGI stack a coding agent reaches for as often as a plain stdlib server.
agentcell login
agentcell deploy --cell fastapi-app .Open the URL agentcell deploy prints, then visit /widgets for the HTML page, or use the JSON
API directly:
curl https://<cell>.agentcell.cloud/api/widgets
curl -X POST https://<cell>.agentcell.cloud/api/widgets -H 'content-type: application/json' -d '{"name":"gear"}'GET / returns the marker line every sample here serves, as plain text — built by hand with
PlainTextResponse rather than FastAPI's default JSON response. GET /api/widgets and
POST /api/widgets are a small JSON API over an in-memory list, validated with a Pydantic model;
GET /api/widgets/{id} returns one or 404s. GET /widgets renders the same list as an HTML page.
The widget list lives in memory only — this sample's point is FastAPI's request/response and
validation shape, not the volume contract (see notes-sqlite for that, with a real database at
/data/notes.db).
Only marker: AGENTCELL_SAMPLE_MARKER if the deploy set one, else a value this app generates
itself the first time it starts, kept so a restart can be told from a redeploy.
Point your agent's MCP client at AgentCell:
{"mcpServers":{"agentcell":{"command":"agentcell","args":["mcp"]}}}See https://agentcell.dev/docs/for-ai-agents.md for the full tool reference.