red

Cut a production release for red and verify it is healthy in production. Use when the user wants to tag and publish a release, monitor the GitHub Release workflow, confirm `red.computer` is healthy after deploy, or reason about rollback for a bad production release.

Full lifecycle for repo daemons — create, seed memory, run, and refine via interactive resume. Use when adding a new `*.daemon.md` audit agent, tightening an existing one, seeding a daemon's initial memory with tracked facts from authority files, or verifying a new daemon reasons correctly before CI rotation.

Debug a red preview deployment on the shared Hetzner box. Use when a preview URL, deploy, remote service, or per-PR stack needs investigation over SSH, or when you need to find where preview services, logs, caddy routes, env files, and compose state live on the host.

Bootstrap new repo experiments with the red conventions for folder layout, runtime config, testing, and compose-based e2e.

Use when debugging red request flow, verifying wide-events ingestion, inspecting MinIO raw or rollup data, checking request_id propagation, or quickly tracing why a request did or did not materialize into a canonical rollup.