name: git-release
description: "Use when pushing changes to production: the mutation loop, running gates locally, pushing to trunk, and reverting failed revisions."
The loop: reproduce -> fix root cause -> run npm run check and the relevant
test tier and read output -> commit with a why -> push to trunk. CI repeats
every binding gate; a push without local checks is sloppy, not blocked.
After a failed deployment: diagnose from the failure wake and CI logs, then
push a revert whose message names the failed revision id (that reconciles the
deployment barrier). Use deployment_rollback only for a diagnosed semantic
failure of the current revision; it restores exactly the previous one, once.
Your worktree ff-only syncs on start; keep it clean or commit deliberately.
---
name: git-release
description: "Use when pushing changes to production: the mutation loop, running gates locally, pushing to trunk, and reverting failed revisions."
---
The loop: reproduce -> fix root cause -> run `npm run check` and the relevant
test tier and read output -> commit with a why -> push to `trunk`. CI repeats
every binding gate; a push without local checks is sloppy, not blocked.
After a failed deployment: diagnose from the failure wake and CI logs, then
push a revert whose message names the failed revision id (that reconciles the
deployment barrier). Use `deployment_rollback` only for a diagnosed semantic
failure of the current revision; it restores exactly the previous one, once.
Your worktree ff-only syncs on start; keep it clean or commit deliberately.