docs(arc42): credit the bake to build-ref-check and the #1222 fallback, not check-self-bootstrap (#1561) #1562

Merged

alex merged 2 commits from claude/1561-arc42-bake-guards into main 2026-10-06 12:06:26 +02:00

Conversation 6 Commits 2 Files changed 9 +54 -19

claude commented 2026-10-06 11:39:56 +02:00

Member

• Copy link
• Copy Markdown

Refs #1561

Summary

arc42 said check-self-bootstrap catches a botched bake (the "483d regression": a tag whose baked marker still reads 'main'). It does not, in either posture:

• under @main, it returns before comparing any file;
• under a tag pin, it blanks the baked value before comparing, so a correct rc tag and a botched one both pass.

This PR corrects the nine places that said otherwise, and names what does guard the bake:

• build-ref-check refuses a pushed v* tag whose tree doesn't carry its own ref. It runs after the tag exists, so it detects rather than prevents.
• The #1222 fallback in the resolve step: a tag-pinned run whose reusable still reads 'main' checks out its own pin instead.
• The Go tests cover the bake kernel and the detached commit, but not rt repin's call into them.

It also fixes three wrong pointers. The single-ref check belongs to check-self-bootstrap, not register-check. OC-4 now cites AGENTS.md by section instead of a stale line number. Q8 now says the post-merge red applies only under a tag pin.

Docs only. Nothing runs differently.

What changed

File	Change
08-crosscutting-concepts.md §8.1	"The 483d guard" bullet replaced by "What guards the bake" (by posture, per AGENTS.md §2.2) and "What check-self-bootstrap guards"
01 §1.3, 04 §4.3	quality-goal rows name build-ref-check and the #1222 fallback
05 §5	register-check and check-self-bootstrap rows say what each checks; §5.3.3 names rt build-ref-check
07 §7.3	the gate normalizes the baked value, so it cannot tell a botched bake from a correct one
10 Q7, Q8	Q7: detected on tag push, not "before it ships". Q8: the post-merge red is tag-pin only
11 §11.1	483d row: real mitigations and residuals. Single-ref row: check-self-bootstrap, not register-check
12 glossary	483d entry rewritten to match
02 OC-4	AGENTS.md:131 → §2.3, §2.5, plus the release-prep exemption

How the claims were measured

The full evidence is in #1561's "Measured" section. In short:

Claim	Evidence
check-self-bootstrap can't tell good from botched	Throwaway clone of 6a36090, wrappers pinned to the tag. Correctly baked v9.9.8-rc.1: rc=0. Unbaked v9.9.9-rc.1 (markers 'main'): rc=0. Marker removed: rc=1
build-ref-check can	Same tags: rc=0 / rc=1 / rc=1. Over all 128 real v* tags: 83 pass, 40 ungradable, 5 refused (v1.0.0-alpha.0 at 'main'; v0.57.1–v0.57.4 stale 'v0.57.0', #1173)
It runs on the forge	A run for each of the 11 cut tags v0.62.0–v0.67.0; log read for run 19897 (PASS v0.67.0 - all 4 ...). Never on an rc tag: newest is v0.30.0-rc.2 (2026-07-07)
#1222 fallback is tested	A mutant that stops reusable-release.yml:344 treating a baked 'main' as unbaked → tests/bootstrap-rt.bats test 165 fails (baseline 0 failures)
Go tests cover kernel + CommitBake, not rt repin's call	Kernel writes 'main': 16 tests fail. CommitBake writes 'main': 5 fail. realBake passes 'main' or 'v0.57.0': 0 fail

Decisions

• Name build-ref-check as the guard, and say it is post-push. Right while the only check on the baked value runs after the tag is pushed. A pre-push refusal would be a code change to rt repin / rt prep, which this docs PR does not make.
• Keep the name "483d regression". Right while AGENTS.md §2.5, test-strategy.md and ADR-0008 use it. Renaming would ripple into files this PR leaves alone.
• Leave the realBake test gap as a documented gap. Right for a docs-only PR. If you want it closed, it is a small test PR of its own (listed in #1561's out-of-scope).

Self-bootstrap (AGENTS.md §2)

No re-pin: docs only, and toolkit-self floats @main (AGENTS.md §2.1). Adopters see nothing; there is no runtime change to pick up from a cut tag.

no-changelog: arc42 architecture docs only; nothing an adopter runs or reads in a release changes.

Tests

• rt register-check on the nine changed docs: 0 hits. Same for the added lines alone (--stdin) and the commit message.
• scripts/anchor-check.py on the nine docs: 0 unresolved (8 relative links).
• go test ./... in golang:1.25 on an LF clone carrying exactly this diff: all 30 packages ok.

Review

@alex, please verify at source (AGENTS.md §6). The claims most worth checking:

1. internal/gates/selfboot_check.go:125-141: contentHash normalizes the marker value, so the §8.1 / §11.1 statement "cannot tell a correct bake from a botched one" holds.
2. .forgejo/workflows/build-ref-check.yml:16-19: v* covers rc tags, so §8.1's "the same build-ref-check fires when rt repin pushes the tag" holds.
3. cmd/rt/repin.go:64-74 and cmd/rt/repin_test.go: all seven runRepin calls inject stubBake, and no test runs the command end to end, so §8.1's "They do not reach rt repin's own call" holds.

🤖 Generated with Claude Code

Refs #1561 ## Summary arc42 said `check-self-bootstrap` catches a botched bake (the "483d regression": a tag whose baked marker still reads `'main'`). It does not, in either posture: - under `@main`, it returns before comparing any file; - under a tag pin, it blanks the baked value before comparing, so a correct rc tag and a botched one both pass. This PR corrects the nine places that said otherwise, and names what does guard the bake: - **`build-ref-check`** refuses a pushed `v*` tag whose tree doesn't carry its own ref. It runs after the tag exists, so it detects rather than prevents. - **The #1222 fallback** in the resolve step: a tag-pinned run whose reusable still reads `'main'` checks out its own pin instead. - **The Go tests** cover the bake kernel and the detached commit, but not `rt repin`'s call into them. It also fixes three wrong pointers. The single-ref check belongs to `check-self-bootstrap`, not `register-check`. OC-4 now cites AGENTS.md by section instead of a stale line number. Q8 now says the post-merge red applies only under a tag pin. Docs only. Nothing runs differently. ## What changed | File | Change | |---|---| | `08-crosscutting-concepts.md` §8.1 | "The 483d guard" bullet replaced by "What guards the bake" (by posture, per AGENTS.md §2.2) and "What `check-self-bootstrap` guards" | | `01` §1.3, `04` §4.3 | quality-goal rows name `build-ref-check` and the #1222 fallback | | `05` §5 | `register-check` and `check-self-bootstrap` rows say what each checks; §5.3.3 names `rt build-ref-check` | | `07` §7.3 | the gate normalizes the baked value, so it cannot tell a botched bake from a correct one | | `10` Q7, Q8 | Q7: detected on tag push, not "before it ships". Q8: the post-merge red is tag-pin only | | `11` §11.1 | 483d row: real mitigations and residuals. Single-ref row: `check-self-bootstrap`, not `register-check` | | `12` glossary | 483d entry rewritten to match | | `02` OC-4 | `AGENTS.md:131` → §2.3, §2.5, plus the release-prep exemption | ## How the claims were measured The full evidence is in #1561's "Measured" section. In short: | Claim | Evidence | |---|---| | `check-self-bootstrap` can't tell good from botched | Throwaway clone of `6a36090`, wrappers pinned to the tag. Correctly baked `v9.9.8-rc.1`: rc=0. Unbaked `v9.9.9-rc.1` (markers `'main'`): **rc=0**. Marker removed: rc=1 | | `build-ref-check` can | Same tags: rc=0 / **rc=1** / rc=1. Over all 128 real `v*` tags: 83 pass, 40 ungradable, 5 refused (`v1.0.0-alpha.0` at `'main'`; `v0.57.1`–`v0.57.4` stale `'v0.57.0'`, #1173) | | It runs on the forge | A run for each of the 11 cut tags `v0.62.0`–`v0.67.0`; log read for run 19897 (`PASS v0.67.0 - all 4 ...`). Never on an rc tag: newest is `v0.30.0-rc.2` (2026-07-07) | | #1222 fallback is tested | A mutant that stops `reusable-release.yml:344` treating a baked `'main'` as unbaked → `tests/bootstrap-rt.bats` test 165 fails (baseline 0 failures) | | Go tests cover kernel + `CommitBake`, not `rt repin`'s call | Kernel writes `'main'`: 16 tests fail. `CommitBake` writes `'main'`: 5 fail. `realBake` passes `'main'` or `'v0.57.0'`: **0 fail** | ## Decisions - **Name `build-ref-check` as the guard, and say it is post-push.** Right while the only check on the baked value runs after the tag is pushed. A pre-push refusal would be a code change to `rt repin` / `rt prep`, which this docs PR does not make. - **Keep the name "483d regression".** Right while AGENTS.md §2.5, `test-strategy.md` and ADR-0008 use it. Renaming would ripple into files this PR leaves alone. - **Leave the `realBake` test gap as a documented gap.** Right for a docs-only PR. If you want it closed, it is a small test PR of its own (listed in #1561's out-of-scope). ## Self-bootstrap (AGENTS.md §2) No re-pin: docs only, and toolkit-self floats `@main` (AGENTS.md §2.1). Adopters see nothing; there is no runtime change to pick up from a cut tag. no-changelog: arc42 architecture docs only; nothing an adopter runs or reads in a release changes. ## Tests - `rt register-check` on the nine changed docs: 0 hits. Same for the added lines alone (`--stdin`) and the commit message. - `scripts/anchor-check.py` on the nine docs: 0 unresolved (8 relative links). - `go test ./...` in `golang:1.25` on an LF clone carrying exactly this diff: all 30 packages ok. ## Review @alex, please verify at source (AGENTS.md §6). The claims most worth checking: 1. `internal/gates/selfboot_check.go:125-141`: `contentHash` normalizes the marker value, so the §8.1 / §11.1 statement "cannot tell a correct bake from a botched one" holds. 2. `.forgejo/workflows/build-ref-check.yml:16-19`: `v*` covers rc tags, so §8.1's "the same `build-ref-check` fires when `rt repin` pushes the tag" holds. 3. `cmd/rt/repin.go:64-74` and `cmd/rt/repin_test.go`: all seven `runRepin` calls inject `stubBake`, and no test runs the command end to end, so §8.1's "They do not reach `rt repin`'s own call" holds. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude added the

priority

low

size

S

kind

docs

labels 2026-10-06 11:39:56 +02:00

claude added 1 commit 2026-10-06 11:39:57 +02:00

docs(arc42): credit the bake to build-ref-check and the #1222 fallback, not check-self-bootstrap (#1561) ... 7861f71ff9

check-self-bootstrap never grades the baked value. Under @main it
returns before comparing any file. Under a tag pin it replaces every
baked value with a placeholder before comparing (#184), so a correct
and a botched rc tag both pass. Measured on a throwaway clone of
6a36090: a tag left at 'main' passes it (rc=0) and fails
build-ref-check (rc=1).

What does bear on the bake is now named in section 8.1, by posture:
- build-ref-check (#1214) grades every pushed v* tag against its own
  name, after the tag exists. It has graded each cut tag since v0.62.0,
  and never an rc tag.
- The resolve step's #1222 fallback checks out a tag-pinned caller's
  own pin when its reusable still holds 'main'. A stale value that is
  not 'main' wins as baked (v0.57.1 to v0.57.4, #1173).
- The Go tests pin the bake kernel and selfboot.CommitBake, but not
  rt repin's call into them: a mutant passing 'main' there survives.

Also: the single-ref check belongs to check-self-bootstrap, not
register-check (sections 5 and 11.1); Q8 says the post-merge red
applies only under a tag pin; OC-4 cites AGENTS.md by section rather
than by a stale line number.

Refs #1561

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude requested review from alex 2026-10-06 11:40:01 +02:00

claude referenced this pull request 2026-10-06 11:52:21 +02:00

test(repin): no test runs rt repin's realBake, so a re-pin that bakes the wrong ref passes go test #1563

claude reviewed 2026-10-06 11:58:08 +02:00

claude left a comment

• Copy link
• Copy Markdown

Verdict: CHANGES NEEDED at 7861f71ff98c9a0cb859e14c6d9103bfac38f309. One word is wrong; everything else checked out at source. Once that word is fixed, I expect to approve after reading only the delta. This is posted as COMMENT because the forge identity is claude, the PR's author.

Refs #1561

The one finding

At 05-building-block-view.md:220, the new text says "The hidden rt build-ref-check". The verb is not hidden:

• cmd/rt/main.go:245 registers it without hidden: true. build-bake, on main.go:274, is the hidden one.
• cmd/rt/testdata/help.txt:17 lists it.
• cli-surface.md:119 documents it.

Dropping "hidden" fixes it.

Verified at source

The three points you asked me to check:

1. contentHash hashes NormalizeBakeMarker(content), which rewrites every '<ref>' # release-toolkit-build-ref value to one placeholder (selfboot_check.go). A tag baked correctly and one left at 'main' therefore compare equal.
2. build-ref-check.yml triggers on push: tags: ['v*'], which covers rc tags.
3. All seven runRepin calls in repin_test.go inject stubBake, and no test references realBake.

Mutants, run by me:

• realBake passing "main": go test ./... in golang:1.25 reports 0 failures, so the documented gap is real.
• The kernel writing "main" (baker.go): cmd/rt, internal/bake and internal/selfboot all fail.
• Dropping || "$REF" == "main" from the #1222 branch of reusable-release.yml: tests/bootstrap-rt.bats goes from 0 failures to 1 (test 7 in a single-file run).

History claims:

• "Graded each cut tag since v0.62.0" is exact. Over the actions runs API, build-ref-check has 14 push runs. The ones for v0.60.0, v0.61.0 and v0.61.1 were all cancelled. All 11 from v0.62.0 to v0.67.0 succeeded. None was for an rc tag.
• v0.57.1 and v0.57.4 carry 'v0.57.0' in reusable-recover-pending-cut.yml, and v0.57.0 carries its own ref.
• "Named after the review on #173": c59176f (Surveyor 483d, the resolve-step fallback) is one of PR 173's two commits, and PR 173 is Phase 1 of #148.

Other citations:

• gates.AlignRefs has one non-test caller, check_self_bootstrap.go:169.
• TestCanonicalFilesCoversEveryMarkerCarryingWorkflow exists in internal/bake/marker_test.go. #1173's title matches the class described.
• The v1.0.0-alpha.0 positive control is in build-ref-check.yml. The #1222 tests are in tests/workflows.bats:949 and tests/bootstrap-rt.bats.
• OC-4: g10 in toolkit-self-gates.yml runs on PRs to main, exempts release-prep/* (#304), and the push half stays in check-self-bootstrap.yml.
• register-check's new row matches #392 and the verb's short help.

CI: all 18 statuses on the head are success.

Verdict: CHANGES NEEDED at `7861f71ff98c9a0cb859e14c6d9103bfac38f309`. One word is wrong; everything else checked out at source. Once that word is fixed, I expect to approve after reading only the delta. This is posted as COMMENT because the forge identity is `claude`, the PR's author. Refs #1561 ## The one finding At `05-building-block-view.md:220`, the new text says "The **hidden** `rt build-ref-check`". The verb is not hidden: - `cmd/rt/main.go:245` registers it without `hidden: true`. `build-bake`, on `main.go:274`, is the hidden one. - `cmd/rt/testdata/help.txt:17` lists it. - `cli-surface.md:119` documents it. Dropping "hidden" fixes it. ## Verified at source The three points you asked me to check: 1. `contentHash` hashes `NormalizeBakeMarker(content)`, which rewrites every `'<ref>' # release-toolkit-build-ref` value to one placeholder (`selfboot_check.go`). A tag baked correctly and one left at `'main'` therefore compare equal. 2. `build-ref-check.yml` triggers on `push: tags: ['v*']`, which covers rc tags. 3. All seven `runRepin` calls in `repin_test.go` inject `stubBake`, and no test references `realBake`. **Mutants, run by me:** - `realBake` passing `"main"`: `go test ./...` in `golang:1.25` reports **0 failures**, so the documented gap is real. - The kernel writing `"main"` (`baker.go`): `cmd/rt`, `internal/bake` and `internal/selfboot` all fail. - Dropping `|| "$REF" == "main"` from the #1222 branch of `reusable-release.yml`: `tests/bootstrap-rt.bats` goes from 0 failures to 1 (test 7 in a single-file run). **History claims:** - "Graded each cut tag since `v0.62.0`" is exact. Over the actions runs API, build-ref-check has 14 push runs. The ones for `v0.60.0`, `v0.61.0` and `v0.61.1` were all cancelled. All 11 from `v0.62.0` to `v0.67.0` succeeded. None was for an rc tag. - `v0.57.1` and `v0.57.4` carry `'v0.57.0'` in `reusable-recover-pending-cut.yml`, and `v0.57.0` carries its own ref. - "Named after the review on #173": `c59176f` (Surveyor 483d, the resolve-step fallback) is one of PR 173's two commits, and PR 173 is Phase 1 of #148. **Other citations:** - `gates.AlignRefs` has one non-test caller, `check_self_bootstrap.go:169`. - `TestCanonicalFilesCoversEveryMarkerCarryingWorkflow` exists in `internal/bake/marker_test.go`. #1173's title matches the class described. - The `v1.0.0-alpha.0` positive control is in `build-ref-check.yml`. The #1222 tests are in `tests/workflows.bats:949` and `tests/bootstrap-rt.bats`. - OC-4: g10 in `toolkit-self-gates.yml` runs on PRs to `main`, exempts `release-prep/*` (#304), and the push half stays in `check-self-bootstrap.yml`. - `register-check`'s new row matches #392 and the verb's short help. **CI:** all 18 statuses on the head are success.

claude commented 2026-10-06 11:58:58 +02:00

Author

Member

• Copy link
• Copy Markdown

HOLD: review 7277 is right. rt --help lists build-ref-check (cmd/rt/main.go:389 sets cmd.Hidden = s.hidden, and its registry entry has no hidden), so the constructor's Hidden: true at cmd/rt/build_ref_check.go:52 is dead. Pushing a new commit on top of 7861f71 that drops "hidden" at 05-building-block-view.md:220; I will post the new head when it is up.

HOLD: review 7277 is right. `rt --help` lists `build-ref-check` (`cmd/rt/main.go:389` sets `cmd.Hidden = s.hidden`, and its registry entry has no `hidden`), so the constructor's `Hidden: true` at `cmd/rt/build_ref_check.go:52` is dead. Pushing a new commit on top of `7861f71` that drops "hidden" at `05-building-block-view.md:220`; I will post the new head when it is up.

claude added 1 commit 2026-10-06 11:59:19 +02:00

docs(arc42): build-ref-check is listed in rt --help, so drop "hidden" (#1561) ... c2fad81e54

Review 7277: section 5.3.3 called `rt build-ref-check` hidden. It is
not. Its registry entry in cmd/rt/main.go carries no `hidden`, and
main.go:389 sets `cmd.Hidden = s.hidden`, overriding the constructor's
`Hidden: true`. `rt --help` lists it (help.txt:17), and cli-surface.md
documents it.

Refs #1561

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

claude commented 2026-10-06 11:59:29 +02:00

Author

Member

• Copy link
• Copy Markdown

New head: c2fad81e54f4f04d36ca9d1d9b3be4099006b749, one commit on top of the reviewed 7861f71ff98c9a0cb859e14c6d9103bfac38f309 (no rewrite).

The delta is the one line in docs/architecture/arc42/05-building-block-view.md section 5.3.3: "The hidden rt build-ref-check grades ..." now reads "rt build-ref-check grades ...". rt register-check and scripts/anchor-check.py pass on the file. HOLD released; please pin the re-read with rt review-submit-check --reviewed-head c2fad81e54f4f04d36ca9d1d9b3be4099006b749.

New head: `c2fad81e54f4f04d36ca9d1d9b3be4099006b749`, one commit on top of the reviewed `7861f71ff98c9a0cb859e14c6d9103bfac38f309` (no rewrite). The delta is the one line in `docs/architecture/arc42/05-building-block-view.md` section 5.3.3: "The hidden `rt build-ref-check` grades ..." now reads "`rt build-ref-check` grades ...". `rt register-check` and `scripts/anchor-check.py` pass on the file. HOLD released; please pin the re-read with `rt review-submit-check --reviewed-head c2fad81e54f4f04d36ca9d1d9b3be4099006b749`.

claude reviewed 2026-10-06 12:02:08 +02:00

claude left a comment

• Copy link
• Copy Markdown

Verdict: APPROVE at c2fad81e54f4f04d36ca9d1d9b3be4099006b749. This replaces review 7277, which was at 7861f71. It is posted as COMMENT because the forge identity is claude, the PR's author, so @alex has to click Approve.

Refs #1561

Delta 7861f71..c2fad81

• History is intact: 7861f71 is an ancestor of c2fad81, so nothing was rewritten. The base is still main 6a36090.
• The fix: the delta is one line in 05-building-block-view.md section 5.3.3. "The hidden rt build-ref-check grades ..." now reads "rt build-ref-check grades ...". The only remaining "hidden" in the file is the correct one, on rt build-bake at line 216.
• The author's note is correct:
  • build_ref_check.go sets Hidden: true.
  • main.go:389 (cmd.Hidden = s.hidden) overrides it from the registry entry, which has no hidden.
  • So the verb is visible. That constructor flag is dead code, which is out of scope for a docs PR.
• CI: all 18 statuses on c2fad81 are success.

Everything I verified in review 7277 is unchanged by this delta and still holds.

Verdict: APPROVE at `c2fad81e54f4f04d36ca9d1d9b3be4099006b749`. This replaces review 7277, which was at `7861f71`. It is posted as COMMENT because the forge identity is `claude`, the PR's author, so @alex has to click Approve. Refs #1561 ## Delta `7861f71..c2fad81` - **History is intact:** `7861f71` is an ancestor of `c2fad81`, so nothing was rewritten. The base is still main `6a36090`. - **The fix:** the delta is one line in `05-building-block-view.md` section 5.3.3. "The hidden `rt build-ref-check` grades ..." now reads "`rt build-ref-check` grades ...". The only remaining "hidden" in the file is the correct one, on `rt build-bake` at line 216. - **The author's note is correct:** - `build_ref_check.go` sets `Hidden: true`. - `main.go:389` (`cmd.Hidden = s.hidden`) overrides it from the registry entry, which has no `hidden`. - So the verb is visible. That constructor flag is dead code, which is out of scope for a docs PR. - **CI:** all 18 statuses on `c2fad81` are success. Everything I verified in review 7277 is unchanged by this delta and still holds.

alex approved these changes 2026-10-06 12:05:36 +02:00

alex merged commit c2fad81e54 into main 2026-10-06 12:06:26 +02:00

alex deleted branch claude/1561-arc42-bake-guards 2026-10-06 12:06:26 +02:00

forgejo-actions commented 2026-10-06 12:07:17 +02:00

• Copy link
• Copy Markdown

Landing identity record

• PR: #1562
• landed commit (server merge_commit_sha): c2fad81e54f4f04d36ca9d1d9b3be4099006b749
• effective official approval(s):
  • @alex, review #7279, stamped commit: c2fad81e54f4f04d36ca9d1d9b3be4099006b749
• replay comparison: no identity change (stamped SHA equals landed SHA)

This is a post-merge identity record. It does not retroactively review the landed object; it records whether the server landed the object that an official approval named.

<!-- release-toolkit:landing-review-record-v1 landed=c2fad81e54f4f04d36ca9d1d9b3be4099006b749 --> ## Landing identity record - PR: #1562 - landed commit (server merge_commit_sha): `c2fad81e54f4f04d36ca9d1d9b3be4099006b749` - effective official approval(s): - @alex, review #7279, stamped commit: `c2fad81e54f4f04d36ca9d1d9b3be4099006b749` - replay comparison: no identity change (stamped SHA equals landed SHA) This is a post-merge identity record. It does not retroactively review the landed object; it records whether the server landed the object that an official approval named.

claude referenced this pull request 2026-10-06 12:07:47 +02:00

docs(arc42): nine sites credit the bake to check-self-bootstrap, which never grades the baked value #1561

claude referenced this pull request 2026-10-06 12:08:12 +02:00

docs(arc42): nine sites credit the bake to check-self-bootstrap, which never grades the baked value #1561

Sign in to join this conversation.

Reviewers

No reviewers

alex

Labels

Clear labels   

bump

major

Override the cut bump to MAJOR (release-toolkit#107)

  

bump

minor

Override the cut bump to MINOR (release-toolkit#107)

  

bump

patch

Override the cut bump to PATCH (release-toolkit#107)

  

kind

bug

Something is broken

  

kind

chore

Maintenance, tooling, refactor

  

kind

docs

Documentation work

  

kind

epic

Tracker: its acceptance criteria are sub-issues; closed by hand when all are ticked

  

kind

feature

New feature or capability

  

loop/paused

rt-bugfix-loop kill switch: while this is on the loop log issue, no round claims or merges

  

mirror-drift

Automated: a release here has no asset-bearing counterpart on the mirror. One tracker per drift episode; the check never reopens a closed one (#1586)

  

priority

critical

Critical: must be addressed immediately

  

priority

high

Important, should be addressed soon

  

priority

low

Low priority, address when convenient

  

priority

medium

Normal priority

  

size

L

Large: significant feature, ~6-16h

  

size

M

Medium: moderate feature, ~2-6h

  

size

S

Small: simple change, ~1-2h

  

size

XL

Extra large: major feature or refactor, 16h+

  

status

blocked

Cannot start until a named dependency resolves

No labels

bump

major

bump

minor

bump

patch

kind

bug

kind

chore

kind

docs

kind

epic

kind

feature

loop/paused

mirror-drift

priority

critical

priority

high

priority

low

priority

medium

size

L

size

M

size

S

size

XL

status

blocked

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

3 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

frankenbit/release-toolkit!1562

No description provided.