feat(work-items): document the re-decompose (rerouting) flow in decompose - #3016
Conversation
…pose Absorb the course's rerouting recipe into /work-items:decompose as a documented usage pattern of existing seam verbs — no new capability: - Five-step flow: close unimplemented children not-planned (provider- mechanic close with a superseding-direction comment; claimed items are coordinated with, not closed from under their holder), keep implemented children untouched, re-interview/edit the spec where it lives, regenerate remaining slices via the normal draft -> approve -> publish steps (create-item --parent --blocked-by), continue on the updated frontier. - Doctrine stated once: tickets are disposable, the spec is editable — slices are projections of the spec and are re-projected, never patched. - Bounds: post-ship wrongness is a new spec, never a patch to a closed container; small drift is an ordinary body edit, not a reroute. - Works against both spec homes: topic Brief and tracker container body. - Lane E verdict recorded ADOPTED in docs/upstream/aihero-shipping-course.md. - work-items 0.36.1. Closes #2949. Refs #2933. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CnzwTKoTa6xNY7iyEzMYpm
|
Claude finished @kyle-sexton's task in 3m 12s —— View job Claude finished the code review
SummaryThis is a clean, internally consistent documentation PR. I traced every seam-verb and cross-reference claim the new
One finding worth a look, posted inline on Note: the PR body's markdownlint/jq/test-suite verification claims are author-reported; I did not independently re-run them (no need to — I checked the actual diff for markdownlint-relevant issues by hand and traced all seam-verb claims against the source). |
|
Claude finished @kyle-sexton's task in 47s —— View job Security review — skipped (no security-relevant surface)
Skip gate triggered: all four changed files are documentation/config — |
|
Last security-reviewed head: |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0ef65a6b71
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Review findings on the no-container variant: (1) step 1 cited a PLAN.md published-slice list the publish step never writes — reconstruct the slice set instead via a provider search for open items citing the topic slug (the Parent provenance line every slice carries), user-confirmed before closing; (2) step 5 routed Brief-only continuation through /work-items:ship, which is a container router whose no-container branch loops back to decompose — a Brief-only journey continues via /work-items:work or the report's frontier ordering. Refs #2949. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CnzwTKoTa6xNY7iyEzMYpm
|
Claude finished @kyle-sexton's task in 44s —— View job Security review — skipped (no security-relevant surface)
Skip gate triggered. Since the last reviewed head ( All four files changed in the full PR diff remain documentation/config: |
|
Claude has reviewed this PR 2 times. The lane skips further automatic reviews after 5; deleting this comment resets the count. |
Review suggestion: the container-lifecycle flow has eval coverage (case 6) and the re-decompose flow carries comparable correctness-critical constraints. Case 7 grades: obsoleted unimplemented slices closed not-planned with a superseding-direction comment, claimed in-flight slice never closed from under its holder, merged children untouched, spec edited before regeneration, approval gate held on regenerated slices, post-ship wrongness treated as a new spec. Refs #2949. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CnzwTKoTa6xNY7iyEzMYpm
|
Claude finished @kyle-sexton's task in 49s —— View job Security review — skipped (no security-relevant surface)
Skip gate triggered: all five changed files are documentation/config — No code, no scripts, no GitHub Actions workflows, no authorization/access-control |
|
Claude finished @kyle-sexton's task in 2m 3s —— View job Claude finished the code review
SummaryRe-reviewed after the two fixup commits (
No new findings on this diff — both open threads from the prior rounds are correctly closed out, and the eval addition doesn't introduce anything questionable. Note: I re-derived the diff via |
Closes #2949
Summary
Absorbs the course's rerouting recipe (Lane E of spec container #2933) into
/work-items:decomposeas a documented re-decompose flow — a usage pattern of existing seam verbs, not a new capability or skill./work-items:ship's existing "slices no longer fit the spec" route now lands on an owned, documented flow.Fix
plugins/work-items/skills/decompose/SKILL.md— new## Re-decompose (rerouting)section covering the five steps with seam verbs named: (1) close unimplemented children as not-planned (provider-mechanic close through the bound adapter, one-line comment linking the superseding direction; claimed items are coordinated with — or routed to/work-items:track auditon a stale lease — never closed from under their holder), (2) keep implemented children untouched, (3) re-interview/edit the spec where it lives (container body under the spec-on-tracker model, or the topic Brief when no container exists), (4) regenerate remaining slices through the normal draft → approve → publish steps (create-item --parent --blocked-by, mandatory approval gate), (5) continue on the updated frontier. The disposable-tickets/editable-spec doctrine is stated once; the "when NOT to reroute" bounds cover post-ship wrongness (new spec, never a patch to a closed container) and small drift (ordinary body edit). Frontmatter description gains re-decompose trigger phrases.docs/upstream/aihero-shipping-course.md— Lane E verdict row flipped to ADOPTED; new## Lane E (#2949)section records the per-concept verdicts.plugins/work-items/CHANGELOG.md/plugin.json— work-items 0.36.1.Acceptance criteria of #2949: re-decompose section with seam verbs named ✔; doctrine stated once ✔; works against both spec homes (topic Brief and tracker container) ✔.
Verification
markdownlint-cli2over the three changed markdown files: 0 issues.jq emptyon the bumpedplugin.json: valid.plugins/work-items/tests/no-hardcoded-priority-scheme.test.sh: PASS.plugins/work-items/tools/work-item-tracker/work-item-tracker.test.sh: 6/6 suites, all PASS (32 cases).Related
Refs #2933 (spec container — Lane E), #2934 (Lane A spec-on-tracker model the flow builds on), #3010 (the
/work-items:shiprouter whose re-slice route lands here).🤖 Generated with Claude Code
https://claude.ai/code/session_01CnzwTKoTa6xNY7iyEzMYpm
Generated by Claude Code