Intake triage skips every issue: repin the author-id gate and add dispatch intake

Problem

intake-triage.yml here calls the reusable lane at v0.45.0, whose author gate (melodic-software/ci-workflows#755) admits a human only when author_association is OWNER or MEMBER. The webhook payload reports CONTRIBUTOR for the org owner, whose membership is private, so every intake run since 2026-10-17 22:27 UTC skipped the triage job behind a green run. Root cause and the reusable fix: melodic-software/ci-workflows#772 (PRs melodic-software/ci-workflows#773 and melodic-software/ci-workflows#774).

Separately, the lane sweep (#7857) files needs-triage issues with the job's GITHUB_TOKEN, which starts no issues run, so those issues never get intake either.

Work here

After a ci-workflows release that contains feat(skill-quality,playbooks): precompute-opportunity check and authoring guidance #773 (and feat(adhd): add ADHD output-shaping plugin (reauthor of i-have-adhd) #774), repin the caller to it.
Fill allowed-author-ids from .github/standards/trusted-actors/trusted-actors.json (the kind: human ids), emitted by the born-triaged job as an output, not hardcoded.
Add workflow_dispatch with an issue-number input; the born-triaged job reads the issue by that number; pass issue-number through and list the dispatching bot's id in allowed-bot-sender-ids.
Decide skipped-label. The only existing fit is needs-triage, which intake-label-needs-triage already owns (it skips issues with a priority: label); labels are managed in github-iac Labels.cs, so a new label means a github-iac change. Recommendation: pass none; the skipped job's notice and summary make the skip visible.
Process rule in docs/conventions/pr-pipeline/README.md: a PR changing a lane that lists a "Not verified" item that could disable the lane does not merge until that item is checked live, and its post-merge live check is a tracked issue.

Live check after the repin merges

Open an issue as kyle-sexton and confirm the claude-intake-triage-status job runs (not skipped); open one from a non-listed account or read a gated-out run and confirm claude-intake-triage-skipped names the reason; dispatch the workflow for one sweep-filed issue and confirm it triages.

Proposed work class: work-class: scoped
Basis: automated intake triage; the lanes App applied this class.

Size: Medium
Complexity: Medium
Uncertainty: Knowable
Flags: none
Blocked by: none

Understood: intake triage skips human-filed issues because the author gate admits only OWNER or MEMBER, and the owner appears as CONTRIBUTOR. Sweep-filed issues also never trigger intake. The work is to:

repin the caller to a ci-workflows release that contains the fix;
emit the trusted human ids from the born-triaged job;
add a workflow_dispatch path with an issue-number input;
decide skipped-label (the issue recommends none);
add the PR-pipeline process rule.

Still to confirm: the repin needs a ci-workflows release containing #773 and #774, which sits outside this repository. The live checks listed in the issue come after the repin merges. No likely duplicates found among the open issues.

_{Automated triage, applied by the lanes App.}

Status, 2026-10-18:

The reusable fix landed: ci-workflows#773 and ci-workflows#774 merged and shipped in v0.49.0 (ff8cd36321423de961381e608e4a31563e74adb3).
The caller repin is #7924, a draft. It is blocked: runner-policy.mjs rejects intake-triage.yml@ff8cd36 because no reviewed contract exists for it ("auto-approval declined: inputs changed"). A standards feat(runner-policy) PR has to register that contract, adding issue-number, allowed-author-ids and allowed-bot-sender-ids to the v0.45.0 entry, and the owner has to sign it off. The standards sync has to land here before fix(intake): admit listed humans by id and triage dispatched issues #7924 can go green.
The process rule is docs(pr-pipeline): hold a lane PR on a not-verified item that can stop the lane #7888.

Remaining live checks, after #7924 merges. No test issues are to be filed; use issues as they come in.

Opened path: the next issue opened by kyle-sexton starts claude-intake-triage-status (not skipped), and the triage comment and labels land.
Skipped path: an issue from an unlisted account, or a gated-out run, runs claude-intake-triage-skipped. Its notice and summary name the reason, and with skipped-label unset no label is applied.
Dispatch path: the lane sweep's GITHUB_TOKEN dispatch (actor github-actions[bot], id 41898282) for a sweep-filed issue starts the run, and the run reads the issue from the API. Still unverified until then: whether allowed_bots in claude-code-action admits github-actions[bot] on workflow_dispatch, and the github.actor a GITHUB_TOKEN dispatch reports. Both were disclosed as unverified on ci-workflows#774.
Dispatch by a person: gh workflow run intake-triage.yml -f issue-number=<N> by kyle-sexton triages issue N. Run it only on an issue that has no triage comment yet.

Live checks after #7924 (merged 08:05:55Z):

Owner-opened issue gets intake: runs 39118063544 and 39118060864 (issues event, born-triaged success, claude-intake-triage-status success, skipped job skipped).
Born-triaged issue is skipped on the opened path: run 39116589749 (skip=true reason=born-triaged).
Owner dispatch from main reaches the gate with the dispatched issue: run 39118376214 (workflow_dispatch, -f issue-number=8083, born-triaged success, skip=true reason=born-triaged, intake skipped). That proves the trigger, the issue lookup and the gate; a dispatch on an untriaged issue (to confirm the triage job itself on dispatch) is still to be seen.
Non-listed author gets the skipped-job notice: not yet seen.
github-actions[bot] dispatch triages (whether claude-code-action allowed_bots accepts it): not yet seen; first exercised by the self-healing sweep.

Posted by Claude Code on the owner's behalf.

Live-check status, 2026-10-19 (no change to the two open boxes):

Non-listed author gets the skipped-job notice: still not seen. The last 100 intake-triage runs were all opened by listed authors (98 by kyle-sexton, 2 by the lanes App). Both lanes-App runs (39258911124, 39255595461) ran claude-intake-triage-status successfully with the skipped job skipped, so a listed bot author is triaged on the opened path.
github-actions[bot] dispatch triages: still not seen, and it can't be seen until an issue qualifies. The hourly triage sweep (maintenance-audit-lane-runs.yml, triage-sweep job) runs on schedule and succeeds (latest 39294303334). It dispatches only for an issue that has no intake comment, no triage or status label, is over an hour old, and was opened by a listed author (triage-sweep.mjs:31-46). Intake has reached every such issue since 2026-10-18, including through the 15:49-17:02Z Actions incident, so the sweep has had nothing to dispatch. The last workflow_dispatch run of intake-triage is still 39118376214.

Both are rare paths. This issue can stay open until one of them happens on its own.

Posted by Claude Code on the owner's behalf.

Status 2026-10-21: the github-actions[bot] dispatch path is proven. In run 39443417878 (2026-10-20 14:53Z, workflow_dispatch, actor and triggering actor github-actions[bot]), claude-intake-triage succeeded and the skipped-notice job was skipped. One case is still open: a non-listed author gets the skipped notice. Per the owner, this stays open until that case is seen.

Released for re-triage under the owner's 2026-10-24 rule: an issue stops for a person only for an intent or vision question, a security call, a host or attended run, or spend outside prepaid credit. The hold label came off and intake triage runs again now; it re-applies a hold, with a numbered question, when one of those applies.

To hold this issue again, add needs-human (or status: needs-decision) and say why in a comment.

Refinement. A read-only run explored this repository, searched its issues and pull requests, and read the allowed docs to draft what follows. On an issue a listed human filed, it counts as the issue's spec once posted, while unedited; it is still model text, so check it.

Intended outcome: Close the issue once the one remaining live check passes: an issue from a non-listed author runs the skipped-notice job; the repin, trusted-ids output, dispatch input and bot-id wiring are already merged.

Related work

fix(intake): admit listed humans by id and triage dispatched issues #7924 (fixed-by): Merged 2026-10-18: admits listed humans by id and triages dispatched issues, covering items 1-4 of the body.
feat(pr-pipeline): add an hourly sweep that re-dispatches intake triage #8180 (related): Added the hourly sweep that re-dispatches intake triage; the github-actions[bot] dispatch path is proven by run 39443417878.
chore(ci): pin ci-workflows v0.62.0 #9140 (related): Later ci-workflows pin bump; the caller is now on a newer pin than this issue asked for.

Open questions

Items 1-4 are merged and live-checked. Should this stay open only for the non-listed-author skipped-notice check, or close with that check moved to a separate tracking issue?
RECOMMENDED: Keep it open only for that check; an unattended run cannot produce a non-listed author event, so no lane run can finish it. Basis: comment 2026-10-21 (owner: stays open until that case is seen); .github/workflows/intake-triage.yml:199-206
Item 5, the process rule in docs/conventions/pr-pipeline/README.md, is not in the README (no 'Not verified' text). Comments say #7888 owns it. Is it dropped from this issue?
RECOMMENDED: Yes, track it only under #7888. Basis: comment 2026-10-18 03:17 (process rule is #7888); grep of docs/conventions/pr-pipeline/README.md found no such rule

Sources read

context.json comments 2026-10-18 to 2026-10-21: repin merged via #7924, bot dispatch path proven, only non-listed-author notice check open
.github/workflows/intake-triage.yml:199-206: reusable lane called with issue-number, allowed-author-ids from born-triaged human-ids output, allowed-bot-sender-ids from bot-ids output
docs/conventions/pr-pipeline/README.md: no Not verified rule found

_{Automated refinement by the lanes App. Answer with a comment whose first line is **Decision recorded.**.}

The work-resolve-issue lane stopped on its usage limit, not on this issue. It keeps agent-ready, and work-unblock starts it again after 2026-10-25T16:00:00Z.

Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/40151023945

Re-triage. This is not an owner stop. The work is done, and one live check waits for an event no lane can produce.

Items 1-4 are merged: fix(intake): admit listed humans by id and triage dispatched issues #7924 (6498194), and the caller is now pinned at ci-workflows v0.65.15 (intake-triage.yml:244). Item 5 is docs(pr-pipeline): hold a lane PR on a not-verified item that can stop the lane #7888, which is closed.
Live checks proven: an owner-opened issue (run 39118063544), the born-triaged skip (run 39116589749), an owner dispatch (run 39118376214), and a github-actions[bot] dispatch (run 39443417878).
The wiring for the last path is covered statically. ci-workflows' intake-triage-gate.test.cjs at the pinned SHA has "an unlisted author waits for a person, OWNER and MEMBER included" and "a gated-out opened issue runs the skipped job, which may apply a label".
Not yet seen live: an issue search for authors other than the owner and the two Apps since 2026-10-18 returns 0.

Per the owner (comment), this stays open until that case is seen. Labels: status: confirmed. agent-ready comes off, because no lane run can open an issue from an unlisted account.

An issue opened by an unlisted account runs claude-intake-triage-skipped, and its notice and summary name the reason. Link the run, then close this issue.

Routed to attended: the one open check needs an issue opened by a GitHub account that is not on the listed-humans list, and no lane or host run has such a login. Opening one test issue from another account, linking its claude-intake-triage-skipped run here and closing both finishes this issue. Agents decided this on evidence under standing direction §1.

Closing: owner decision, 2026-11-03. The owner declined to make an account just to run this check: "I'm not going to go create a fake account for this".

What is already proven:

Live: an owner-opened issue (run 39118063544), the born-triaged skip (run 39116589749), an owner dispatch (run 39118376214), and a github-actions[bot] dispatch (run 39443417878).
In tests, not live: the unlisted-author path. ci-workflows' intake-triage-gate.test.cjs at the pinned SHA covers it with "an unlisted author waits for a person, OWNER and MEMBER included" and "a gated-out opened issue runs the skipped job, which may apply a label".

The first issue opened by an account outside the listed humans exercises that path for real. Reopen this issue if its claude-intake-triage-skipped run does not name the reason.

Navigation Menu

Repository navigation

Intake triage skips every issue: repin the author-id gate and add dispatch intake #7886

Problem

Work here

Live check after the repin merges

3 remaining items

Metadata

Assignees

Labels

Type

Fields

Projects

Milestone

Relationships

Development

Intake triage skips every issue: repin the author-id gate and add dispatch intake #7886

Description

Problem

Work here

Live check after the repin merges

Activity

github-actions commented on Oct 18, 2026

kyle-sexton commented on Oct 18, 2026

kyle-sexton commented on Oct 18, 2026

kyle-sexton commented on Oct 19, 2026

kyle-sexton commented on Oct 21, 2026

kyle-sexton commented on Oct 25, 2026

3 remaining items

melodic-automation-lanes commented on Oct 25, 2026

melodic-automation-lanes commented on Oct 25, 2026

kyle-sexton commented on Oct 28, 2026

kyle-sexton commented on Oct 31, 2026

kyle-sexton commented on Nov 3, 2026

Metadata

Metadata

Assignees

Labels

Type

Fields

Projects

Milestone

Relationships

Development

Issue actions