Cull Release Recover
Principle
Derive truth from external evidence, then resume the first incomplete safe phase.
Never erase immutable release history to make local state look consistent.
Diagnose
- Resolve the Cull checkout.
- Run
npm run release:cull -- resume --version "$VERSION" --json and parse
exactly one JSON envelope. Treat local release state as a cache.
- Authenticate and compare the prepared commit, remote annotated tag and peeled
commit, Actions run, exact asset inventory, public release, updater metadata,
provenance, and Homebrew cask/evidence.
- Classify the next action exactly as
rerun-check, prepare-new-version,
watch-build, verify-artifact, publish-verified-artifacts,
promote-homebrew, or prepare-patch-plan.
- Execute only the safe resumable phase the user already authorized. Do not
repeat a completed signed build or replace verified artifacts.
Burned-tag recovery (STALE_RELEASE_SOURCE / post-publish failure)
When a pushed tag's workflow fails and the tag is immutable (protected tags),
never move it. Follow the v0.5.0 / v0.6.1 / v0.6.2 precedent:
- Record the failure:
state fail --version "$VERSION" --code GATE_FAILED|PUBLISH_FAILED|... --evidence-json '{...}'.
- Fix the underlying cause on main through a normal PR (never inside the tag).
- Prepare the next version from current main. Pick the next deeper number
(failed 0.6.0 → 0.6.1 → 0.6.2). Curated notes still follow the user-facing
norm: cover the full delta since the last public release, plain language.
- The recovery release PR may carry small non-owned commits (smoke records,
doc fixes) — only the release commit is file-scoped.
- Re-run the human smoke gate for the new version (fresh build, fresh
docs/releases/<v>-smoke.md, fresh closed bead with
external_ref=cull-release-<version>-smoke in the landing worktree's bd).
- Re-anchor the record to the release PR merge commit before tagging (see
cull-release-publish).
- The
resume command derives state from the local cache and can lag external
reality (e.g. it may say watch-signed-build for a workflow that already
failed). Authenticate the workflow run itself before trusting it.
A post-publish verification failure creates or updates the stable P0 bd incident,
blocks later release checks, and returns prepare-patch-plan. Prepare the patch
plan but never publish that patch implicitly.
Never delete or move a tag or release. Never force-push, clobber assets, reset
unrelated work, bypass a gate, clean cull.db, or treat missing evidence as
success.
1---2name: cull-release-recover3description: Use when a Cull release is stuck, inconsistent, failed in a workflow or artifact gate, missing Homebrew promotion, failed after publication, or needs a safe recovery or patch plan.4metadata:5 source: https://skillmd.com/skills/glebis/cull-release-recover6---78# Cull Release Recover910## Principle1112Derive truth from external evidence, then resume the first incomplete safe phase.13Never erase immutable release history to make local state look consistent.1415## Diagnose16171. Resolve the Cull checkout.182. Run `npm run release:cull -- resume --version "$VERSION" --json` and parse19 exactly one JSON envelope. Treat local release state as a cache.203. Authenticate and compare the prepared commit, remote annotated tag and peeled21 commit, Actions run, exact asset inventory, public release, updater metadata,22 provenance, and Homebrew cask/evidence.234. Classify the next action exactly as `rerun-check`, `prepare-new-version`,24 `watch-build`, `verify-artifact`, `publish-verified-artifacts`,25 `promote-homebrew`, or `prepare-patch-plan`.265. Execute only the safe resumable phase the user already authorized. Do not27 repeat a completed signed build or replace verified artifacts.2829## Burned-tag recovery (STALE_RELEASE_SOURCE / post-publish failure)3031When a pushed tag's workflow fails and the tag is immutable (protected tags),32never move it. Follow the v0.5.0 / v0.6.1 / v0.6.2 precedent:33341. Record the failure: `state fail --version "$VERSION" --code GATE_FAILED|PUBLISH_FAILED|... --evidence-json '{...}'`.352. Fix the underlying cause on main through a normal PR (never inside the tag).363. Prepare the next version from current main. Pick the next deeper number37 (failed 0.6.0 → 0.6.1 → 0.6.2). Curated notes still follow the user-facing38 norm: cover the full delta since the last public release, plain language.394. The recovery release PR may carry small non-owned commits (smoke records,40 doc fixes) — only the release *commit* is file-scoped.415. Re-run the human smoke gate for the new version (fresh build, fresh42 `docs/releases/<v>-smoke.md`, fresh closed bead with43 `external_ref=cull-release-<version>-smoke` in the **landing worktree's** bd).446. Re-anchor the record to the release PR merge commit before tagging (see45 cull-release-publish).467. The `resume` command derives state from the local cache and can lag external47 reality (e.g. it may say `watch-signed-build` for a workflow that already48 failed). Authenticate the workflow run itself before trusting it.4950A post-publish verification failure creates or updates the stable P0 bd incident,51blocks later release checks, and returns `prepare-patch-plan`. Prepare the patch52plan but never publish that patch implicitly.5354Never delete or move a tag or release. Never force-push, clobber assets, reset55unrelated work, bypass a gate, clean `cull.db`, or treat missing evidence as56success.