← Registry
Loop Official

Boot and Probe

Green tests are not a running app — build the production image, boot it against a fresh throwaway database, probe the real routes and contracts, fix, and re-boot until everything answers.

Unit tests and a passing build cannot see module-load side effects, missing env vars, migrations failing on a blank database, or routes that 500 only at runtime. Each iteration builds the production artifact, boots it against a disposable environment (a fresh database container on a side port — never your dev instance), probes the real routes and public contracts, and fixes the first runtime failure before rebooting. Distinct from independent-verifier-pass (static gates: build, lint, tests): this proves RUNTIME behavior on a real boot, before merge rather than after deploy.

Loop

Goal
the production image builds, boots cleanly against a fresh throwaway database, and every probed route and contract returns its expected response
Exit when
the image builds and boots against the throwaway database and every probed route returns its expected response

Check between iterations

docker build -t app-smoke . && docker run -d --rm -p 3000:3000 --env-file .env.smoke app-smoke && sleep 5 && curl -fsS http://localhost:3000/
$

Kickoff prompt

Start the "Boot and Probe" loop. Goal: the production image builds, boots cleanly against a fresh throwaway database, and every probed route and contract returns its expected response Max iterations: 6 Between iterations run: docker build -t app-smoke . && docker run -d --rm -p 3000:3000 --env-file .env.smoke app-smoke && sleep 5 && curl -fsS http://localhost:3000/ Exit when: the image builds and boots against the throwaway database and every probed route returns its expected response Step 1: Spin up a fresh disposable database on a side port, build the production image, boot it against that database, and probe the home route plus every public contract the change touches.