Deployment

PyFlue is a Python package, so deployment is centered on Python runtimes, containers, and CI jobs.

The closest PyFlue equivalent to a JavaScript server target is the Docker and FastAPI target. In a workspace, runtime builds write a dist/ directory with server.py, manifest.json, provider config, and any container files needed by the selected platform.

Supported Targets

Target	Status	Command	Notes
New Build System
Uvicorn	Implemented	pyflue build --target uvicorn	FastAPI server with SSE/webhook support.
Lambda	Implemented	pyflue build --target lambda	AWS Lambda with Mangum adapter.
Cloud Run	Implemented	pyflue build --target cloudrun	Google Cloud Run with gunicorn.
Docker	Implemented	pyflue build --target docker	General Python container.
CI/CD
GitHub Actions	Implemented	pyflue build --target github-actions	Generates .github/workflows/pyflue-agent.yml.
GitLab CI/CD	Implemented	pyflue build --target gitlab-ci	Generates .gitlab-ci.yml.
Platforms
Railway	Implemented	pyflue build --target railway	Generates Docker/FastAPI artifacts and railway.json.
Render	Implemented	pyflue build --target render	Generates Docker/FastAPI artifacts and render.yaml.
Fly.io	Implemented	pyflue build --target fly	Generates Docker/FastAPI artifacts and fly.toml.
Vercel	Implemented	pyflue build --target vercel	Generates Python app artifacts and vercel.json.
Netlify	Implemented	pyflue build --target netlify	Generates Python app artifacts and netlify.toml.
Cloudflare Containers	Beta	pyflue build --target cloudflare	Generates Docker/FastAPI artifacts, worker.ts, wrangler.jsonc, and package.json.

Uvicorn/FastAPI

Generate a FastAPI server with built-in development support:

pyflue build --target uvicorn

This writes:

dist/server.py
dist/requirements.txt
dist/manifest.json

Run the server:

cd dist
pip install -r requirements.txt
python server.py

The server exposes: - GET /health - Health check - GET /agents - List available agents - POST /agents/{name}/{agent_id} - Run an agent

AWS Lambda

Generate an AWS Lambda handler with Mangum adapter:

pyflue build --target lambda

This writes:

dist/main.py
dist/requirements.txt
dist/manifest.json

Deploy to Lambda using the AWS CLI or SAM. The handler is handler.

Google Cloud Run

Generate a Cloud Run optimized container:

pyflue build --target cloudrun

This writes:

dist/server.py
dist/Dockerfile
dist/requirements.txt
dist/cloudbuild.yaml
dist/manifest.json

Deploy using:

gcloud run deploy --source .

Docker/FastAPI

Generate the default Python web artifacts:

pyflue build --target docker

This writes:

dist/Dockerfile
dist/server.py
dist/requirements.txt
dist/manifest.json

The generated server exposes the PyFlue server:

curl http://localhost:8000/prompt/default \
  -H "Content-Type: application/json" \
  -d '{"payload": {"prompt": "Review this project"}}'

File-based agents are exposed under /agents/{name}/{agent_id}.

GitHub Actions

Generate a workflow:

pyflue build --target github-actions

This writes:

.github/workflows/pyflue-agent.yml

The workflow is manual by default. Add provider keys as repository secrets, then run it from the GitHub Actions tab.

GitLab CI/CD

Generate a pipeline file:

pyflue build --target gitlab-ci

This writes:

.gitlab-ci.yml

The generated job is designed for manual web pipelines. Add provider keys as masked CI/CD variables.

Railway

Generate Railway files:

pyflue build --target railway

This writes:

dist/server.py
dist/Dockerfile
dist/requirements.txt
dist/railway.json
dist/manifest.json

Deploy the project with Railway's GitHub integration or CLI.

pyflue deploy --target railway

When the Railway CLI is installed and authenticated, PyFlue runs railway up.

Render

Generate Render files:

pyflue build --target render

This writes:

dist/server.py
dist/Dockerfile
dist/requirements.txt
dist/render.yaml
dist/manifest.json

Create a new Blueprint in Render and point it at the repository.

Fly.io

Generate Fly.io files:

pyflue build --target fly

This writes:

dist/server.py
dist/Dockerfile
dist/requirements.txt
dist/fly.toml
dist/manifest.json

Review the generated app name and region before deploying with fly deploy.

pyflue deploy --target fly

When the Fly.io CLI is installed and authenticated, PyFlue runs fly deploy.

Cloudflare Containers

Generate Cloudflare Containers files:

pyflue build --target cloudflare

This target generates:

dist/server.py
dist/Dockerfile
dist/requirements.txt
dist/worker.ts
dist/wrangler.jsonc
dist/package.json
dist/manifest.json

Cloudflare Containers are currently a beta Workers feature and require a Workers Paid plan. wrangler deploy builds and pushes the Docker image, so Docker must be running locally during deployment.

Deploy with Wrangler:

npm install
npm run deploy

Vercel

Generate Vercel files:

pyflue build --target vercel

This writes:

dist/server.py
dist/requirements.txt
dist/vercel.json
dist/manifest.json

Review the generated route config before deploying.

pyflue deploy --target vercel

When the Vercel CLI is installed and authenticated, PyFlue runs vercel deploy.

Netlify

Generate Netlify files:

pyflue build --target netlify

This writes:

dist/server.py
dist/requirements.txt
dist/netlify.toml
dist/netlify/functions/pyflue.py
dist/manifest.json

Review the generated function settings before deploying.

pyflue deploy --target netlify

When the Netlify CLI is installed and authenticated, PyFlue runs netlify deploy.