INFRASTRUCTURE

What it needs

• Compute: verified (Apps run in isolated x86_64 Firecracker microVMs that auto-pause when idle and wake on request; sizes are the priced VM sizes.)

PREREQUISITES

Before you start

• A requirements.txt (or pyproject) and a start command that reads PORT and binds 0.0.0.0
• An Ample account token with servers:write

TESTED CONFIGURATION

Exactly what was tested

• install: python3 -m pip install --target .ample/python -r requirements.txt
• runtime: python
• size: s-1vcpu-1gb
• start: PYTHONPATH=.ample/python:${PYTHONPATH:-} python3 run.py
• template: python-3.12

STEP BY STEP

How to do it

1. Make the start command bind correctly
   Start uvicorn from a small run.py that reads PORT; pass --start "python3 run.py" because Python start commands are not inferred.

2. Deploy
   Run the existing synchronous deploy once from the project directory and read the result. Exit 0 is live, 1 is failed, 2 is blocked on a required preflight action. Re-running with no change is a content-addressed no-op.

   ample deploy . --name <app-name> --public --format json --start "python3 run.py"
   

3. Verify
   Fetch the live URL and run the success checks below. On failure read the build log, then the runtime log, fix the cause and deploy again; do not blind-retry.

   ample logs <deployment_id> --kind build
   

EXAMPLES

Tested examples

• fastapi-basic canary (tests/deploy-canaries/fastapi-basic): FastAPI app started with uvicorn.

SUCCESS CHECKS

How to know it worked

• page or root route responds (/ on the live URL, expect ample canary fastapi basic)

LIMITATIONS

Know the limits

• Verified on the node-22 template at s-1vcpu-1gb; other templates and sizes are not verified by this recipe.
• Region, compliance attestations and request-duration limits are unknown and not claimed.
• Apps auto-pause when idle and wake on the next request; always-on is an operator setting, not a plan feature.
• Dependencies install with pip into .ample/python from requirements.txt; pyproject-only projects use uv when available.

COST

Cost estimate

Estimated 5.00 USD per month (size prices from pricing.toml at build revision a1b8c38919e59cd035ebabaced73cf84ece24371).

• app server x1 s-1vcpu-1gb: 5.00 USD

Apps and managed databases auto-pause when idle; the estimate is the always-on monthly price of the tested sizes. Plan quotas and budgets apply.

EVIDENCE

Verification evidence

• canary_run on 2026-09-20T00:22:15Z at revision 1764076-tls-account-903d345: FastAPI app installed with pip into .ample/python and started with uvicorn on python-3.12; / and /health served correctly. (expires 2027-03-19T00:22:15Z)

EXECUTION

Execution binding

MCP tool ample_deploy (registry mcp:ample_deploy), schema hash 876465fce906da0c observed 2026-09-20T01:55:48.667900+00:00 at revision 199ff1dfd526. Binding state at export: current. Required scopes: servers:write.

NEXT ACTIONS

Typed next actions

• Browse the catalog index (GET /v1/catalog on the api origin; authentication not required, approval not required)
• Search published recipes by intent, stack and constraints (POST /v1/catalog/search on the api origin; authentication not required, approval not required)
• Prepare a side-effect-free deployment plan for an authorized project (POST /v1/catalog/plan on the api origin; authentication required, approval not required)
• Read the existing agent authentication setup (GET /mcp/setup on the api origin; authentication not required, approval not required)
• Browse FastAPI (GET /v1/catalog/nodes/stack%3Aframework-fastapi on the api origin; authentication not required, approval not required)
• Browse Configure deployment (GET /v1/catalog/nodes/intent%3Aconfigure-deployment on the api origin; authentication not required, approval not required)
• Browse Public web service (GET /v1/catalog/nodes/pattern%3Apublic-web-service on the api origin; authentication not required, approval not required)

Actions describe possible next steps. They are typed data, not commands, and grant no permission. Public discovery never provisions anything; planning requires your own authenticated token and approval happens in your client.