feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478) by kyle-sexton · Pull Request #641 · melodic-software/claude-code-plugins

kyle-sexton · 2026-07-20T07:46:12Z

Summary

Absorbs bullets 1–4 of the v4 loop-prompt routing rules into the /work-items:triage skill (plugins/work-items/skills/triage/SKILL.md), so the skill owns them instead of a session prompt — per the ratified operator decision on #478 (ACCEPTED option (a) verbatim). Bullet 5 stays deferred to #459 (pointer only); bullet 6 (wayfind:* label semantics) is cross-repo label policy noted on github-iac#176 and left untouched here.

Fix

Decision-defaulted ready route — "Triage states" now documents three briefed exits (delegable, decision-defaulted, human-gated). A single-fork item whose brief carries a well-grounded RECOMMENDED answer with only a maintainer-vetoable alternative routes to the autonomous-eligible role with status:ready plus a Decision defaulted: X — veto before merge comment. Step 2 carries the routing test; "Apply outcome" adds the matching row. Human-gated is narrowed to genuinely open decisions (open design space, product intent, cross-repo policy), with capability blockers (external access, manual QA) kept as a distinct non-delegability axis.
Cluster-aware routing — "Gather context" adds a cluster-detection cross-reference; "Apply outcome" documents the decision-carrier pattern using the native blocked-by edge and a blocked by #<carrier> decision member comment, with no per-member human-gated label. No new labels.
Multi-surface T1 stub — "Apply outcome" adds a lightweight briefing variant for T1 fixes spanning 3+ surfaces (one-line sites + fix pattern comment, autonomous-eligible role, symbol-level durability naming).
Severity sub-sort — the priority-label step records the finding's self-labeled severity when a rule sets priority: above it. No new labels.

Version bump 0.13.0 → 0.14.0 (read from live origin/main), plus a ## [0.14.0] CHANGELOG entry under plugins/work-items/.

Verification

Acceptance criteria, with the actual text added:

Three briefed exits + narrowed human-gated — "Triage states":

delegable — fully specified with no open decision → autonomous-eligible role (default agent-ready).

decision-defaulted — a single-fork item whose brief carries a well-grounded RECOMMENDED answer with only a maintainer-vetoable (reversible) alternative → autonomous-eligible role with status:ready, plus a Decision defaulted: X — veto before merge comment.

human-gated — reserved for a genuinely open decision (open design space, product intent, or cross-repo policy), or work that cannot be delegated for a capability reason (external access, manual QA) → human-gated role (default needs-human).

Decision-defaulted route + verbatim comment in step 2 and Apply outcome — step 2 "Recommend category + state":

apply the routing test: is the alternative reversible/maintainer-vetoable (→ decision-defaulted: autonomous-eligible role + status:ready, recorded with a Decision defaulted: X — veto before merge comment) or genuinely open …

Apply-outcome row:

| Briefed, decision-defaulted | … Apply labels + the autonomous-eligible role label (default agent-ready) + status:ready, and post a Decision defaulted: X — veto before merge comment |

The comment string's dash is a U+2014 em dash (verified e2 80 94), matching the ratified decision verbatim.

Cluster detection + decision-carrier pattern, no per-member human-gated — "Gather context":

Cluster detection — … Designate one representative as the decision carrier (human-gated, with the member numbers listed in its body) and link every other member to it via the native blocked-by edge with a blocked by #<carrier> decision comment …

Apply-outcome:

Decision-carrier clusters. … apply human-gated to the carrier only … Each other member instead gets a native blocked-by edge to the carrier plus a blocked by #<carrier> decision comment — never a per-member human-gated label.

T1 multi-surface stub + durability rule — Apply-outcome row:

| Briefed, T1 multi-surface stub | For a trivial (T1) fix spanning 3+ surfaces: in place of a full brief, post a one-line sites + fix pattern comment and apply the autonomous-eligible role label (default agent-ready). The brief durability rule still holds — name sites by interface / symbol / domain concept, not file paths or line numbers (recommended default: symbol-level naming) |

Severity sub-sort — priority-label step:

when a directive or category rule sets this label above the finding's self-labeled severity, record the original severity in the triage comment (e.g. priority set to pX by <rule>; reporter severity: <sev>) …

Bullet 5 pointer only — Direction gate, appended sentence, no fix text:

Formalizing this as the autonomous-mode contract — codifying that standing-lane rules constitute direction — is tracked in #459.

wayfind:* unchanged — grep wayfind over the skill returns no matches before or after this change.

Checks run locally on the changed files: markdownlint-cli2 → 0 errors; check-skill-portability.sh --paths … SKILL.md → no unexcused coupling tokens; plugin.json parses as valid JSON. The six existing triage evals assert PR-as-item, verify-before-interview, decompose exclusion, raw-marker clearing, and autonomous-lane direction — none assert human-gated-vs-defaulted routing, so these additive changes flip no existing eval.

Closes #478

work-items(triage): step-2 "wait for user direction" gate contradicts the skill's own autonomous/agent triage mode #459 — autonomous-mode contract (bullet 5), deferred; captured here only as a pointer, not fixed.
github-iac#176 — cross-repo complexity/label policy where bullet 6 (wayfind:* label semantics) was noted per the issue's own history; not touched here, not fixed.

🤖 Generated with Claude Code

https://claude.ai/code/session_01KyLj6oaFVTE6xFuoYCC2KC

…ll (#478) Absorb bullets 1-4 of the v4 loop-prompt routing rules into /work-items:triage so the skill owns them instead of a session prompt: - Decision-defaulted ready route: third briefed exit alongside delegable and human-gated; single-fork items with a well-grounded RECOMMENDED answer and a maintainer-vetoable alternative route to autonomous-eligible + status:ready with a "Decision defaulted: X — veto before merge" comment. Routing test added to step 2; human-gated narrowed to genuinely open decisions (or capability blockers). - Cluster-aware routing: cluster-detection cross-reference in Gather context; one decision carrier (human-gated), members linked via native blocked-by edge + "blocked by #<carrier> decision" comment, no per-member human-gated label. - Multi-surface T1 stub: one-line "sites + fix pattern" comment variant for T1 fixes spanning 3+ surfaces; symbol-level durability naming. - Severity sub-sort: record self-labeled severity when a rule sets priority above it. Bullet 5 (autonomous-mode contract) captured only as a pointer to #459. Bullet 6 (wayfind:* semantics) untouched — cross-repo label policy noted on github-iac#176. Version bump 0.13.0 -> 0.14.0. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KyLj6oaFVTE6xFuoYCC2KC

claude · 2026-07-20T07:46:29Z

Claude finished @kyle-sexton's task in 2m 32s —— View job

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

Gather context and read PR diff
Review plugin.json version bump
Review CHANGELOG.md entry
Review SKILL.md routing rule changes
Verify alignment with CLAUDE.md guidelines
Post final review

Overall verdict: approve with minor notes. The changes are well-structured, internally consistent, and correctly implement the four stated routing rules. No blocking issues found. Notes below are observations and one potential ambiguity worth a follow-up issue.

`plugin.json`

plugins/work-items/.claude-plugin/plugin.json

Version bump 0.13.0 → 0.14.0 is correct for additive skill behavior. The description field still doesn't reference the new routing distinctions, but it's already long and this isn't typically where behavior details live. No issue.

`CHANGELOG.md`

plugins/work-items/CHANGELOG.md

Entry is well-structured, follows Keep a Changelog convention (no date for unreleased is acceptable here), and the Added/Changed split is correct — the human-gated narrowing is a changed behavior rather than a new feature, so its placement under ### Changed is right.

`SKILL.md` — routing rule changes

✅ Decision-defaulted route

Line 53–57, line 95, line 124

The three-exit definition (delegable / decision-defaulted / human-gated) is coherent and the routing test in step 2 (reversible/maintainer-vetoable → defaulted; genuinely open → human-gated) is crisp and implementable. The em dash in Decision defaulted: X — veto before merge is correctly U+2014 throughout.

⚠️ Note: `delegable` exit not labeled in the ASCII diagram

Line 59–65

raw → verified → briefed → autonomous-eligible (role label, default agent-ready)
 |        |          ├→ decision-defaulted → …
 |        |          └→ human-gated …

The diagram's straight-through line (briefed → autonomous-eligible) implicitly represents the delegable path, but it isn't labeled as such. Now that all three exits are named in the prose above, a reader scanning only the diagram won't see delegable called out. Low-friction fix: prepend ├→ delegable → on the top line as a branch alongside the other two. Fix this →

⚠️ Note: T1 stub row omits `status:ready`

Line 125

For a trivial (T1) fix spanning 3+ surfaces: in place of a full brief, post a one-line sites + fix pattern comment and apply the autonomous-eligible role label (default agent-ready).

The briefed state definition (state table, line 48) carries status:ready. The delegable and decision-defaulted rows both result in status:ready on the item. The T1 stub row only mentions agent-ready — no status:ready. This could leave a T1 stub item missing the label the autonomous lane typically expects to see alongside agent-ready for frontier pickup. If the omission is intentional (T1 stubs get a lighter hand-off that doesn't mark them ready), the table row should say so explicitly; if it's an oversight, status:ready should be added.

✅ Cluster detection

Line 87, line 135

The forward reference ("applied in step 5") and the step 5 paragraph correctly close the loop. The "never a per-member human-gated label" constraint is clearly stated. One observation: the carrier-selection criterion ("designate one representative") doesn't specify how to pick the carrier when multiple items qualify (e.g., oldest? highest-severity? reporter's item?). This is left to discretion, which may be intentional to keep the rule simple — but if inconsistent triage behavior surfaces in practice, an explicit tiebreaker rule would be easy to add.

✅ Severity sub-sort

Line 94

The format priority set to pX by <rule>; reporter severity: <sev> is specific and unambiguous. Trailing period is absent on this bullet (minor formatting; other step 2 bullets don't have one either, so it's consistent within the step).

✅ Bullet 5 pointer / wayfind:* untouched

Line 102

Deferred pointer is correctly appended as a non-behavioral sentence. wayfind: produces zero matches in the changed SKILL.md as verified.

CLAUDE.md alignment

Repo-agnostic — no hardcoded paths or repo-specific values introduced. ✅
Plugin-form-safe — all plugin-internal references continue to use ${CLAUDE_PLUGIN_ROOT}. ✅
No PII/secrets — skill definition only, no credentials or personal data. ✅
Configurable without editing — no new config scalars; existing userConfig keys unchanged. ✅
Security — this is prose instructions, not executable code. The direction gate in step 2 is preserved; cluster-detection writes (blocked-by edge) are correctly deferred to step 5 and gated behind the direction check. No regressions. ✅

Summary

Two notes worth a follow-up if not addressed before merge: (1) label the delegable branch in the ASCII diagram so all three exits are visible there, and (2) clarify whether the T1 stub outcome intentionally omits status:ready. Neither is blocking — the skill is behaviorally correct and internally consistent as written.

- Label the delegable branch in the Triage-states diagram so all three briefed exits (delegable, decision-defaulted, human-gated) are visible in the diagram, not just in prose. - Make status:ready explicit on the T1 multi-surface stub outcome: the stub replaces the full brief but not the ready-to-work state, matching the other autonomous-eligible outcomes. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KyLj6oaFVTE6xFuoYCC2KC

claude · 2026-07-20T07:53:19Z

Claude finished @kyle-sexton's task in 1m 6s —— View job

Review update — previous notes addressed

Gather context and identify what changed
Read updated SKILL.md to verify review notes were addressed
Verify delegable branch labeled in ASCII diagram
Verify T1 stub row includes status:ready
Post final review verdict

Commit 84e8edd addresses both notes from the prior review. Changes are targeted and correct:

✅ `delegable` branch now labeled in the diagram

SKILL.md L59–65

The straight-through line has been replaced with an explicit three-branch fork:

raw → verified → briefed
 |        |          ├→ delegable → autonomous-eligible (role label, default agent-ready)
 |        |          ├→ decision-defaulted → autonomous-eligible + status:ready + "Decision defaulted: … — veto before merge"
 |        |          └→ human-gated (role label, default needs-human) — briefed for a human

All three exits are now visible in the diagram, matching the prose definition above it. ✅

✅ T1 stub row now includes `status:ready`

SKILL.md L125

The row now reads agent-ready + status:ready, with an explicit rationale: "the stub replaces the full brief but not the ready-to-work state, so the item is picked up like any other autonomous-eligible outcome." This aligns the T1 stub with the delegable and decision-defaulted rows. ✅

No new issues introduced. This PR is ready to merge.

kyle-sexton · 2026-07-20T07:56:04Z

Review classification (work-2 lane, post-green pass)

#	Finding	Classification	Disposition
1	ASCII state diagram left the `delegable` exit unlabeled while the other two branches now carry labels	VALID (low)	Fixed — all three briefed exits (`delegable`, `decision-defaulted`, `human-gated`) now labeled in the diagram
2	T1 multi-surface stub row didn't say whether it carries `status:ready` like the other two autonomous-eligible outcomes	VALID (low)	Resolved — added `status:ready` to the row. The stub replaces the full brief, not the ready-to-work state; matches bullet 3's "still autonomous-eligible" intent from the ratified decision

Both fixed and verified. 17/17 CI green including the review re-run.

claude · 2026-07-20T07:56:23Z

Claude finished @kyle-sexton's task in 1m 10s —— View job

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

Gather context and read changed files at HEAD (84e8edd)
Verify plugin.json version bump
Verify CHANGELOG.md entry
Verify SKILL.md routing rule changes
Confirm prior review notes were resolved
Check CLAUDE.md alignment

This is a third-pass review over the final state of the branch. The two notes from the first review were addressed in commit 84e8edd and confirmed in the second review. This pass re-reads all three changed files at HEAD to confirm the branch is clean.

`plugin.json`

plugins/work-items/.claude-plugin/plugin.json L4

Version 0.14.0 is present and correct. No other fields changed. ✅

`CHANGELOG.md`

plugins/work-items/CHANGELOG.md L6–41

## [0.14.0] entry is present, well-structured, and follows Keep a Changelog. Added/Changed split is correct — the human-gated narrowing is a behavioral change, not a new feature, so its placement under ### Changed is right. All four bullets from the PR description are represented. ✅

`SKILL.md`

Three briefed exits — L53–57: delegable / decision-defaulted / human-gated are defined clearly and consistently. ✅

ASCII diagram — L59–66: all three exits are now explicitly labeled as branches (├→ delegable, ├→ decision-defaulted, └→ human-gated). The first review note is fully resolved. ✅

Routing test — L96: reversible/maintainer-vetoable → decision-defaulted; genuinely open → human-gated. The test is crisp and implementable. ✅

T1 stub row — L126: now carries agent-ready + status:ready with explicit rationale ("the stub replaces the full brief but not the ready-to-work state"). The second review note is fully resolved. ✅

Cluster detection — L88, L136: carrier pattern closes the loop across steps 1 and 5. "Never a per-member human-gated label" is clearly stated. ✅

Severity sub-sort — L95: format priority set to pX by <rule>; reporter severity: <sev> is specific and unambiguous. ✅

Bullet 5 pointer — L103: deferred to #459 as a non-behavioral sentence. ✅

wayfind:* untouched: zero matches in the skill file. ✅

Em dash — Decision defaulted: X — veto before merge uses U+2014 consistently across the three-exit prose, routing test, and apply-outcome table row. ✅

CLAUDE.md alignment

Rule	Status
Repo-agnostic — no hardcoded paths	✅
Plugin-form-safe — `${CLAUDE_PLUGIN_ROOT}` refs preserved	✅
Configurable without editing — no new `userConfig` scalars	✅
No PII / secrets	✅
Versioned — semver bump present	✅
Direction gate preserved, cluster-detection mutations gated to step 5	✅

No issues found. This PR is ready to merge.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 84e8eddafa

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".

…) (#643) ## Summary `/work-items:work` selection had no rule excluding an issue that already has an open linked PR from the pickable frontier. An issue keeps `status: ready` for its entire window with an open PR (from open through merge/auto-close), so a picker had to manually cross-check `gh pr list` to avoid re-picking work already in flight and starting a duplicate branch. This ships the ratified in-repo open-linked-PR selection filter (operator decision, 2026-07-19): an issue with an open linked PR is not pickable. ## Fix - **Selection-time frontier filter (`skills/work/SKILL.md`, Step 1).** After `list-frontier` derives the frontier, tiers 2–3 candidates are additionally filtered to drop any number that has an open PR targeting it for closure — so an in-flight item leaves the pickable set instead of being re-picked. The **closing-keyword linkage** is authoritative (the same `Closes #N` / native-closing-keyword signal `pr-issue-linkage` enforces); an intentional `Refs #N` opt-out does not exclude its issue. The filter **fails open** when the bound provider exposes no PR host (offline `local-markdown` is never a coordination surface and touches no network tool). - **New GitHub adapter mechanic (`tools/work-item-tracker/adapters/github/README.md`, "Open linked PRs").** The provider mechanics stay in the adapter per this repo's seam/adapter separation — the skill core inlines no `gh`. It uses `gh pr list --state open --search "<N> in:body"` (truncation-safe per item, no page-size race) and a closing-keyword `jq` test with an exact-number boundary so `#463` does not match `#4630` / `#1463`. - **Retires the interim heuristic.** The prior "Already-in-flight (interim, retire on `#463`)" bullet in the execute-step staleness pre-check is removed — the frontier itself now excludes in-flight items, which is exactly the retirement that bullet anticipated. Scope-narrowed to the closing-keyword signal (the interim branch-pattern-only match is dropped): a standard-flow PR always carries the gate-enforced `Closes #N`, so keyword-matching is faithful to the ratified "references it as a closing target" scope. - **No seam-contract change.** `CONTRACT.md` and `list-frontier` are untouched; the addition is a README-documented adapter mechanic, not a new seam verb. The durable seam-level in-review state is explicitly NOT built here (see Related). ## Verification The adapter mechanic was exercised against live repo data (read-only). `<N>` substituted per row: ``` issue #435 open-closing-PR: true (open, unassigned, status: ready; open PR #638 closes it) issue #487 open-closing-PR: true (open, unassigned, status: ready; open PR #629 closes it) issue #534 open-closing-PR: true (open, unassigned, status: ready; open PR #634 closes it) issue #463 open-closing-PR: false (no open PR closes it → stays pickable) issue #4630 open-closing-PR: false (exact-number boundary: does not false-match on #463's PR) ``` Each of #435 / #487 / #534 is a genuine frontier-eligible candidate (open, unassigned, `status: ready`) that under the old logic stayed pickable despite an open PR already in flight; the filter returns `true` for exactly those and excludes them, while #463 (no open closing PR) returns `false` and remains pickable. `list-frontier` itself was not run end-to-end here because the `work-items` plugin repo binds no tracker (`.work-item-tracker.json` is a consuming-repo artifact) — the per-number filter, which is the added logic, is what is demonstrated above. Closes #463 ## Related - #463 — this issue (open-linked-PR selection filter). - #416 — planning routes through the tracker seam; argues the durable in-review state belongs at the seam, not a GitHub-only label. This PR does NOT implement that durable state. - #498 — seam read-verb coverage; where a durable in-review / container-scoped read would live. This PR does NOT implement that durable state. Note: PR #641 concurrently bumps `work-items` to `0.14.0` (minor, for #478) and also edits `plugin.json` + `CHANGELOG.md`. This PR is a patch bump to `0.13.1` from the current `origin/main` base (`0.13.0`) — a different version slot, not a race — but whichever merges second will need a mechanical rebase on those two files (the same-plugin concurrency #464 treats as an awareness note, not a block). 🤖 Generated with a Claude Code implementation subagent (issue #463) --------- Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>

…b-v4-rules # Conflicts: # plugins/work-items/.claude-plugin/plugin.json # plugins/work-items/CHANGELOG.md

claude · 2026-07-20T09:07:46Z

Claude finished @kyle-sexton's task in 1m 50s —— View job

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

Gather context and read PR diff
Review plugin.json version bump
Review CHANGELOG.md entry
Review SKILL.md routing rule changes
Verify alignment with CLAUDE.md guidelines
Post final review

This pass covers the current HEAD (1136d09) — a merge of origin/main into the branch that resolved conflicts in plugin.json and CHANGELOG.md. The three changed files from the PR content (ab7d060 + 84e8edd) were already confirmed clean in the two prior review cycles; this pass focuses on whether the merge resolution introduced any issues.

Merge conflict resolution (`1136d09`)

origin/main gained a 0.13.1 fix release (be8d06e — open-linked-PR exclusion from work frontier) while this branch was in review. Two conflicts:

plugin.json — main: 0.13.1; this branch: 0.14.0. Resolution: 0.14.0. ✅ Correct. 0.14.0 is numerically and semantically above 0.13.1 (additive skill behavior warrants a minor bump).

CHANGELOG.md — main gained a ## [0.13.1] entry; this branch added ## [0.14.0]. Resolution: both entries are present in descending-version order (0.14.0 → 0.13.1 → 0.13.0). ✅ Keep-a-Changelog convention preserved.

SKILL.md — not in the conflict list; the merge touched no work-items triage files beyond plugin.json and CHANGELOG. The file at HEAD is identical to 84e8edd (the state already confirmed clean). ✅

Previously confirmed (no regressions)

Check	Status
Three briefed exits (delegable / decision-defaulted / human-gated) defined and consistent	✅
ASCII diagram labels all three exits explicitly	✅
Routing test (reversible → defaulted; genuinely open → human-gated) in step 2	✅
`Decision defaulted: X — veto before merge` em dash is U+2014 throughout	✅
T1 stub row carries `agent-ready` + `status:ready` with rationale	✅
Cluster detection + decision-carrier pattern; no per-member human-gated label	✅
Severity sub-sort format `priority set to pX by <rule>; reporter severity: <sev>`	✅
Bullet 5 pointer only (non-behavioral sentence, deferred to #459)	✅
`wayfind:*` — zero matches in SKILL.md	✅
Repo-agnostic, plugin-form-safe, no PII, configurable without editing	✅

Open Codex finding (#656)

The P2 gap filed by Codex (cluster non-carrier members carry no positive status marker after raw is cleared, so the attention view's "Unlabeled" bucket re-selects them) was triaged as VALID (defer) by the author with the remedy analysis (A/B/C shapes and why A destroys the one-touch property) now living in issue #656. That classification is sound — this is a design decision about the attention-view contract, not a mechanical error in this PR's content. Not blocking.

This PR is ready to merge.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 1136d09840

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".

…sit merge tier The triage lane records a defaulted (maintainer-vetoable) decision only as a 'Decision defaulted: X — veto before merge' issue comment, which a deterministic merge gate cannot see (Codex P1 on #641). Add a tier criterion: the PR's linked issue must carry no unratified decision-default marker. A marker is ratified only by a human maintainer (authorAssociation OWNER/MEMBER) comment strictly after it; reactions are not consulted (the reactions API carries no author association, so a reaction cannot be attributed to a maintainer, and using the operator's self-logins would let pipeline automation clear its own veto — the same #450 attribution-drift hazard). Fail closed: an unratified marker, or a linked issue whose comments cannot be fetched, holds the PR for the human merge-ready list. Marker matching is deliberately loose (over-matching only holds more for the human). Reuses fetch_issue_comments and the shared actor_kind predicate; criterion codified in safety.md. Pass, fall-back, and fetch-error fixtures added. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

kyle-sexton added the automated Opened by automation. label Jul 20, 2026

kyle-sexton mentioned this pull request Jul 20, 2026

ops: loop-lane telemetry — running per-lane status (babysit / triage / work) #502

Open

kyle-sexton mentioned this pull request Jul 20, 2026

fix(work-items): exclude open-linked-PR issues from work frontier (#463) #643

Merged

kyle-sexton marked this pull request as ready for review July 20, 2026 07:56

chatgpt-codex-connector Bot reviewed Jul 20, 2026

View reviewed changes

Comment thread plugins/work-items/skills/triage/SKILL.md

kyle-sexton mentioned this pull request Jul 20, 2026

work-items(triage): parked cluster members re-surface in the attention view (no positive marker) #656

Open

Merge remote-tracking branch 'origin/main' into feat/478-triage-absor…

1136d09

…b-v4-rules # Conflicts: # plugins/work-items/.claude-plugin/plugin.json # plugins/work-items/CHANGELOG.md

chatgpt-codex-connector Bot reviewed Jul 20, 2026

View reviewed changes

Comment thread plugins/work-items/skills/triage/SKILL.md

kyle-sexton mentioned this pull request Jul 20, 2026

feat(source-control): babysit autopilot merge tier — bot-account review + criteria-gated merge #476

Closed

kyle-sexton merged commit e881b3b into main Jul 20, 2026
17 checks passed

kyle-sexton deleted the feat/478-triage-absorb-v4-rules branch July 20, 2026 09:15

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)#641

feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)#641
kyle-sexton merged 3 commits into
mainfrom
feat/478-triage-absorb-v4-rules

kyle-sexton commented Jul 20, 2026

Uh oh!

claude Bot commented Jul 20, 2026 •

edited

Loading

Uh oh!

claude Bot commented Jul 20, 2026 •

edited

Loading

Uh oh!

kyle-sexton commented Jul 20, 2026

Uh oh!

claude Bot commented Jul 20, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

claude Bot commented Jul 20, 2026 •

edited

Loading

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

kyle-sexton commented Jul 20, 2026

Summary

Fix

Verification

Related

Uh oh!

claude Bot commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

plugin.json

CHANGELOG.md

SKILL.md — routing rule changes

✅ Decision-defaulted route

⚠️ Note: delegable exit not labeled in the ASCII diagram

⚠️ Note: T1 stub row omits status:ready

✅ Cluster detection

✅ Severity sub-sort

✅ Bullet 5 pointer / wayfind:* untouched

CLAUDE.md alignment

Summary

Uh oh!

claude Bot commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Review update — previous notes addressed

✅ delegable branch now labeled in the diagram

✅ T1 stub row now includes status:ready

Uh oh!

kyle-sexton commented Jul 20, 2026

Review classification (work-2 lane, post-green pass)

Uh oh!

claude Bot commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

plugin.json

CHANGELOG.md

SKILL.md

CLAUDE.md alignment

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

claude Bot commented Jul 20, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

PR Review — feat(work-items): absorb v4 loop-prompt routing rules into triage skill (#478)

Merge conflict resolution (1136d09)

Previously confirmed (no regressions)

Open Codex finding (#656)

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

claude Bot commented Jul 20, 2026 •

edited

Loading

`plugin.json`

`CHANGELOG.md`

`SKILL.md` — routing rule changes

⚠️ Note: `delegable` exit not labeled in the ASCII diagram

⚠️ Note: T1 stub row omits `status:ready`

claude Bot commented Jul 20, 2026 •

edited

Loading

✅ `delegable` branch now labeled in the diagram

✅ T1 stub row now includes `status:ready`

claude Bot commented Jul 20, 2026 •

edited

Loading

`plugin.json`

`CHANGELOG.md`

`SKILL.md`

claude Bot commented Jul 20, 2026 •

edited

Loading

Merge conflict resolution (`1136d09`)