source-control/pull-request: probe the Monitor timeout cap in -p runs and fix a re-emit on poll 2 #6877

Activity

1. kyle-sexton

   added

   needs-triageNot yet classified. Floor until a type and one priority tier are set.Not yet classified. Floor until a type and one priority tier are set.

   on Oct 10, 2026
2. kyle-sexton

   added the

   Bug

   issue type on Oct 10, 2026
3. github-actions commented on Oct 10, 2026

   

   github-actionsBot

   on Oct 10, 2026 – with GitHub Actions

   Last edited by melodic-automation-lanes

   Contributor

   More actions

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

   Size: Small
   Complexity: Low
   Uncertainty: Knowable
   Flags: none
   Blocked by: none

   Understood: the failed-first-fetch re-emit fix and its test are already done (#8150). What remains is a probe. A -p run makes a Monitor call with timeout_ms: 1800000 and records whether Claude Code clamps or rejects it. The result goes into plugins/source-control/skills/pull-request/reference/monitor.md with the Claude Code version. The probe needs a claude run, which this lane can't do, so it waits for an attended session. Nothing is missing from the issue. I found no likely duplicates.

   _{Automated triage, applied by the lanes App.}
4. github-actions

   added

   needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

   on Oct 10, 2026
5. kyle-sexton

   removed

   needs-triageNot yet classified. Floor until a type and one priority tier are set.Not yet classified. Floor until a type and one priority tier are set.

   on Oct 12, 2026
6. kyle-sexton commented on Oct 13, 2026

   

   kyle-sexton

   on Oct 13, 2026

   MemberAuthor

   More actions

   Owner batch-ratified this admission in an attended session on 2026-10-12: work-class: scoped. The owner held it until the laptop's auto-merge plan (Poteto Workflow) confirmed no phase touches the PR monitor poll code, and it has now confirmed that. Rebase on main before working it.
7. kyle-sexton

   added

   work-class: scopedA briefed fix or small feature; blast radius bounded by the brief, tests exist.A briefed fix or small feature; blast radius bounded by the brief, tests exist.

   agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.Fully specified and briefed; eligible for autonomous pickup from the frontier.

   and removed

   needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

   on Oct 13, 2026
8. kyle-sexton commented on Oct 13, 2026

   

   kyle-sexton

   on Oct 13, 2026

   MemberAuthor

   More actions

   Implementation note from the verification pass (.work/backlog-strategy/verify/v-13.jsonl in the main checkout): changing the baseline file alone does not stop the re-emit. The monitor's check fetch uses 2>/dev/null ... || true, so a failed poll empties the baseline and the next good poll re-emits every finished check. The fix must keep the previous baseline when a fetch fails, and the first arm has no full-iteration snapshot to seed from. Rebase on main first: #7338 changed the monitor.md push paragraph.
9. kyle-sexton

   added

   priority: lowNice-to-have, cosmetic, or speculative; opportunistic.Nice-to-have, cosmetic, or speculative; opportunistic.

   and removed

   agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.Fully specified and briefed; eligible for autonomous pickup from the frontier.

   on Oct 14, 2026
10. kyle-sexton

    added a sub-issue on Oct 18, 2026

    • fix(source-control): keep the PR monitor's check baseline when a check read fails #7986
11. kyle-sexton commented on Oct 18, 2026

    

    kyle-sexton

    on Oct 18, 2026

    MemberAuthor

    More actions

    This was generated by AI during backlog shaping.

    This issue is split into 2 pieces, each sized for one lane PR. Owner ratified scoped (2026-10-13 comment). The poll fix is lane work and its premise holds on origin/main. The -p timeout probe needs a live claude -p run the lane cannot make.

    • filed: fix(source-control): keep the PR monitor's check baseline when a check read fails (fix(source-control): keep the PR monitor's check baseline when a check read fails #7986)
    • not filed: docs(source-control): record how claude -p handles a Monitor timeout_ms of 1800000: paid-run: needs a live claude -p Monitor probe with timeout_ms 1800000

    Filed pieces are sub-issues of this one, and the paced queue admits them to the lane. Unfiled pieces wait on the blocker shown. The owner approved this split in an attended backlog-shaping session on 2026-10-18.
12. kyle-sexton

    added

    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

    on Oct 19, 2026
13. 9 remaining items

    Load more
14. melodic-automation-lanes

    added

    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

    and removed

    admission-deferred

    agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.Fully specified and briefed; eligible for autonomous pickup from the frontier.

    on Oct 25, 2026
15. melodic-automation-lanes commented on Oct 25, 2026

    

    melodic-automation-lanesBot

    on Oct 25, 2026

    Contributor

    More actions

    The work-resolve-issue lane stopped partway through this issue and needs a person.

    Why, in the run's own words:

    The only open acceptance criterion needs a live probe: one Monitor call with timeout_ms 1800000 in a single-prompt claude -p run, plus the Claude Code version it ran on. This lane session has no Monitor tool, and its Bash may run only the lane wrappers, so it cannot run claude --version or a nested claude -p. The docs state the 10-minute cap for -p runs but not whether a larger value is clamped or rejected, so the behavior cannot be recorded from a source. The issue's own 2026-10-18, 2026-10-19 and 2026-10-24 comments already say this probe needs a live claude run on the operator machine. No files were changed.

    What the run says it did not do:

    • Probe a single-prompt claude -p run calling Monitor with timeout_ms 1800000, then record clamp versus reject and the Claude Code version in plugins/source-control/skills/pull-request/reference/monitor.md. This lane has no Monitor tool and cannot run claude, so it needs a person or an operator-machin
    • PR prep (review, build/test, lint): skipped because nothing changed.

    Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/40208783221

    To retry, remove needs-human and add agent-ready again.
16. kyle-sexton

    changed the issue type from

    Task

    to

    Bug

    on Oct 26, 2026
17. kyle-sexton commented on Oct 26, 2026

    

    kyle-sexton

    on Oct 26, 2026

    MemberAuthor

    More actions

    This was generated by AI.

    Routed to the host runner, not the owner. The poll fix shipped (#7986). Left: one claude -p Monitor call with timeout_ms 1800000 and the version it ran on; the issue lane cannot run it (https://github.com/melodic-software/claude-code-plugins/actions/runs/40208783221).

    Under the owner's 2026-10-26 stop rule, a person is needed only for a change in intent or vision, an implementation finding that changes the intent or primary design, or a high or critical security issue; host, attended and eval runs are agent work. So needs-human comes off. It moves to the host-run label, and the host runner takes it.
18. kyle-sexton

    added

    host-runNeeds an unattended run on a host; the host runner takes it, not a person or the issue lane.Needs an unattended run on a host; the host runner takes it, not a person or the issue lane.

    status: confirmedA maintainer confirmed the item is real and in scope; ready to be worked.A maintainer confirmed the item is real and in scope; ready to be worked.

    and removed

    status: readyTriaged, unblocked, and fully specified; eligible to pick up.Triaged, unblocked, and fully specified; eligible to pick up.

    needs-humanHuman-in-the-loop required; autonomous sessions must not resolve items carrying this.Human-in-the-loop required; autonomous sessions must not resolve items carrying this.

    on Oct 27, 2026
19. kyle-sexton commented on Oct 28, 2026

    

    kyle-sexton

    on Oct 28, 2026

    MemberAuthor

    More actions

    Host run. One single-prompt claude -p session made a Monitor call with timeout_ms: 1800000 to see whether Claude Code clamps or rejects the value.

    What ran:

    • claude -p "<one Monitor call: timeout_ms 1800000, command 'echo armed; sleep 1700'; report the result and the deadline notice>" --permission-mode auto --output-format stream-json --verbose. Each stream line was timestamped. Run in a throwaway directory.
    • Claude Code version: 2.1.312.

    Result against the open acceptance criterion ("the monitor guidance states the -p behavior with the probe's Claude Code version"):

    • Clamped, not rejected. The Monitor call went in with timeout_ms: 1800000. The tool result said the watch "expires in 10m", and the structured result carried timeoutMs: 600000.
    • The watch was armed at 14s and its expiry notice ("Monitor expired after 10m with 1 event delivered") came at 614s. The -p session stayed up to receive the notice, then exited 0 at 617s.
    • No guidance change is needed: the arming call and the re-arm-on-deadline rule work as written in a -p run. The behavior and version are recorded in plugins/source-control/skills/pull-request/reference/monitor.md §3.0.1, as a probe pointer record.
    • The other criterion (a failed first fetch does not re-emit finished checks) was already done by fix(source-control): keep the PR monitor's check baseline when a check read fails #8150.

    PR (opened as a draft, not marked ready): #10604

    Target labels: add none; remove host-run.
20. kyle-sexton

    removed

    host-runNeeds an unattended run on a host; the host runner takes it, not a person or the issue lane.Needs an unattended run on a host; the host runner takes it, not a person or the issue lane.

    on Oct 28, 2026
21. kyle-sexton

    added 2 commits that reference this issue on Oct 28, 2026

    docs(source-control): keep the -p Monitor probe behavior behind its p…

    ...

    354ff1c

    docs(source-control): state that -p clamps the Monitor deadline rathe…

    ...

    8b51e35
22. kyle-sexton

    closed this as completedin #10604on Oct 28, 2026