Repository navigation
chore(work-issues): fold the 2026-10-13 run's lessons into claim, verify and ship - #4880
Merged
Merged
Conversation
…ify and ship - claim.md §4: the pre-claim re-check also reads other issues' claims and claim extensions naming the same files, the run's own lanes included (#4857's claim took src/cli/config-loader.ts, which #2100's extension already held). Second occurrence of the backlog row about claims on another issue. - verify.md §8-b: a lane round that overlaps the parent's §8-d probe edits in scratch and copies in only once the parent frees the tree, since the probe needs a clean tree and restores src from HEAD. That handshake had to be invented mid-run. Paid for by compressing §8-i's marker sentence. - ship.md §9: no merge while another lane's integ runs; #4862's merge restarted #4864's batch. Recorded as the fifth occurrence on the backlog's main-moved-under-integ row. - tooling-backlog: first-occurrence rows for the migrate-from-cfn fixture that leaked an SSM Document per run while exiting 0, and for the classifier refusing a dispatch that bulk-deleted other runs' orphans.
- Every skill file keeps its line count (the §8-b addition is paid for by reflowing and tightening §8-b itself); restore `unknown hash "diff"`, `main` and "anyway" that the compression dropped. - ship.md §9: "ANY turn holding for HOURS" no longer reads as narrowed to a fixture-blocked turn; the merge rule cites #4864 only and says "can move". - claim.md §4: point the new clause at the "working on this" claims, and extend the reason clause to a peer's claim. - tooling-backlog: tie the fifth occurrence to the fourth's shape, and mark the classifier row's "other runs' orphans" reading as an inference.
…agent's call Restores the tie-break referent in claim.md and verify.md §8-b's "as soon as" wording, both lost to the line-count reflow.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The
/work-issuesstage-10 retro for the 2026-10-13 run (closed 6 / filed 1, new 1 / folded 0; nonextfiled, so nothing to promote).ship.md§9: "nor a merge during another lane's integ"; fifth occurrence noted on the main-moved-under-integ backlog rowverify.md§8-b, one sentencemigrate-from-cfnexited 0 while leaking one SSM Document per run (#4873)src/cli/config-loader.tsvia a claim extension (#2100 / #4857)claim.md§4 pre-claim re-check; backlog row marked SECOND occurrence and SHIPPEDEvery skill file keeps its line count against
origin/main(claim.md 111, ship.md 227, verify.md 199), paid for by reflowing and tightening the touched paragraphs; bytes grow by 56-76 per file.Tooling Policy 1 exception, for the maintainer to decide: the
verify.md§8-b sentence (a lane round overlapping the §8-d probe edits in scratch until the parent frees the tree) is a FIRST occurrence. It goes into the stage file rather than a backlog row because retro §10-a/§10-b route a step the run had to INVENT to the stage where it fires, and the next overlap would otherwise destroy a lane's uncommittedsrcedits (the probe's restore readsHEAD). If the maintainer prefers the strict reading, it becomes a backlog row.Verification
verify.md§8-c): every path, section and issue number the new text names resolves in this repo.vp check --fix,vp run check,vp run typecheck:test,vp run build,gen:all-matrices+audit:coverage:check+format(tree unchanged),vp test run --maxWorkers=4: 1859 files / 47175 tests passed, rc=0.