Deploy FastAPI on DockHive

01

Deploy with Drone (zero config)

Drone is DockHive's AI deploy agent. It reads the repository, works out the stack, writes the Dockerfile, picks the port and health check and lists the env vars your FastAPI app needs.

1. 1 Push to GitHubPush the project to a GitHub repository. No Dockerfile or deploy config needed.
2. 2 Pick the repoIn the DockHive console, open App Hosting and choose the repository.
3. 3 Deploy with DroneDrone builds it. If the build or health check fails it reads the logs, fixes the build setup and retries, then goes live on <app>.dockhive.app with HTTPS.

Drone saves the working setup, so every push to the repository redeploys automatically. How Drone works

02

Deploy from the terminal

Install the hivepod CLI, sign in once, then run hivepod deploy in the project folder. It uploads the folder (without .git, node_modules, virtualenvs or secrets) and Drone builds it.

Deploying the same folder again updates the same app. Set variables on the way with hivepod deploy --env KEY=value.

03

Deploy from your AI agent (MCP)

The CLI includes an MCP server. Add it to Claude Code, then ask your agent to "deploy this to DockHive". The agent calls the deploy tool and hands back the live URL, or Drone's summary and any missing env vars if something needs your input.

Using Claude Desktop, Cursor or Windsurf? Set up the DockHive MCP server.

04

What DockHive detects for FastAPI

Detection reads your files; it never runs your code. Recognised by a fastapi dependency. These are the defaults; DockHive derives the exact commands from your repository (lockfile, scripts, entry module).

Default port

8000

Install

pip install -r requirements.txt

Start

uvicorn main:app

• Finds your app object (any name = FastAPI(...)) and serves it with uvicorn on 0.0.0.0:$PORT.
• uvicorn[standard] is installed for you when uvicorn is not in your dependencies.
• Port: read from code such as uvicorn.run(..., port=...) when present.
• Recognised by requirements.txt, pyproject.toml, Pipfile, setup.py or manage.py.
• Python version: read from .python-version or runtime.txt (3.9 to 3.13; 3.12 when unset).
• Install: requirements.txt when present; otherwise poetry.lock (Poetry), uv.lock (uv), Pipfile (Pipenv) or pip install . for pyproject.toml/setup.py.
• psycopg2 (not -binary) and mysqlclient get the system packages they need to compile.
• Generated images set PORT, listen on 0.0.0.0:$PORT and run as a non-root user wherever the runtime allows. Nothing is written back to your repository.
• Monorepo? When the repository root isn't an app, DockHive looks one level down (app, web, frontend, backend, server, api, src, site, client first).

05

Environment variables and custom domains

FastAPI on DockHive: FAQ