feat(pr-pipeline): review an oversize main-run diff in chunks instead of holding it for a person #11000

Activity

1. github-actions

   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 29, 2026
2. github-actions commented on Oct 29, 2026

   

   github-actionsBot

   on Oct 29, 2026 – with GitHub Actions

   Last edited by melodic-automation-lanes

   Contributor

   More actions

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

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

   Understood: change pr-review-main.yml so a diff over MAX_DIFF_BYTES is split by file into chunks under the cap and reviewed per chunk. The verdict binds only when every chunk returns a counted verdict, and the highest CRITICAL count wins. A failed or missing chunk fails closed. A single file over the cap still records main-review-stop:too-large, and chunk count and per-chunk bytes are recorded as metrics. The acceptance criteria are concrete and the approach is clear. The change edits a gate workflow, so the security review lane should judge the diff. I found no likely duplicates among the open issues.

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

   added

   priority: mediumReal value, no hard deadline; normal backlog flow.Real value, no hard deadline; normal backlog flow.

   and 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 29, 2026
4. github-actions

   added the

   Feature

   issue type on Oct 29, 2026
5. melodic-automation-lanes

   added

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

   work-class: structuralRefactors, migrations, contract changes; cross-cutting and hard to reverse.Refactors, migrations, contract changes; cross-cutting and hard to reverse.

   admission-deferredAdmitted but waiting for WIP room; the hourly work-unblock sweep labels it agent-ready.Admitted but waiting for WIP room; the hourly work-unblock sweep labels it agent-ready.

   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

   admission-deferredAdmitted but waiting for WIP room; the hourly work-unblock sweep labels it agent-ready.Admitted but waiting for WIP room; the hourly work-unblock sweep labels it agent-ready.

   on Oct 29, 2026
6. melodic-automation-lanes commented on Oct 29, 2026

   

   melodic-automation-lanesBot

   on Oct 29, 2026

   Contributor

   More actions

   This records a re-fire of an earlier refusal, not a new refusal. The lanes kill switch stopped run 40934459461, the work-resolve-issue run that agent-ready started, before it did any work; that run ended at 2026-10-29T18:56:56Z. work-unblock has now removed and added agent-ready back to start a fresh run. A kill-switch refusal is no fault in this issue or the lanes, so it never counts against later re-fires; after limits.limit-stops-per-issue of them since a person last added agent-ready or a run last reported, work-unblock waits longer before each one, up to 24 hours.
7. melodic-automation-lanes

   added

   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

   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 29, 2026
8. melodic-automation-lanes commented on Oct 29, 2026

   

   melodic-automation-lanesBot

   on Oct 29, 2026

   Contributor

   More actions

   This records a re-fire of an earlier refusal, not a new refusal. The lanes kill switch stopped run 40952840274, the work-resolve-issue run that agent-ready started, before it did any work; that run ended at 2026-10-29T20:56:11Z. work-unblock has now removed and added agent-ready back to start a fresh run. A kill-switch refusal is no fault in this issue or the lanes, so it never counts against later re-fires; after limits.limit-stops-per-issue of them since a person last added agent-ready or a run last reported, work-unblock waits longer before each one, up to 24 hours.
9. melodic-automation-lanes

   added

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

   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.

   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 29, 2026
10. melodic-automation-lanes commented on Oct 30, 2026

    

    melodic-automation-lanesBot

    on Oct 30, 2026

    Contributor

    More actions

    The work-resolve-issue lane stopped on this issue (patch-protected-path): the change needs a path the issue lane may not push, which a maintainer session builds by hand.
    It added host-run and removed agent-ready: an agent run on a host builds it, and it waits on no decision from a person.

    In the run's words:

    The run stopped early (patch-protected-path): its change needs a path the issue lane may not push, which a person changes instead. In the run's words: patch-protected-path: .github/workflows/pr-review-main.yml. The size stop (stop too-large when diff.patch exceeds MAX_DIFF_BYTES) is inline shell in the 'Write the diff' step of each of the workflow's three review legs (lines 243-245, 732-734, 1216-1218). The single-pass review is one run-claude-cli step per leg (line 384 and its counterparts). Reviewing the diff in chunks with a per-chunk verdict that binds only when every chunk returns one means changing those steps, so the change needs .github/workflows/pr-review-main.yml, which this lane cannot patch.

    What the run says it did not do:

    • Chunked review in .github/workflows/pr-review-main.yml (per-chunk run-claude-cli review across the three legs, binding only when every chunk returns a counted verdict, max CRITICAL across chunks, fail closed on a missing chunk, keeping the too-large stop only for a single file over the cap): not sta
    • Chunk-count and per-chunk-bytes metrics (§9) and the fixture tests in .github/scripts/pr-review-main/workflow.test.mjs: not started, because they depend on the workflow change
    • Independent security review of the gate-workflow change: not run, because no change was made
    • tidy, review, verify and checks pipeline steps: skipped, because the run stopped before any edit

    Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/40975913532

    To hand it back to the lane, remove host-run and add agent-ready again.
11. kyle-sexton

    added 2 commits that reference this issue on Oct 30, 2026

    fix(pr-pipeline): keep each file whole in a main-review chunk and tak…

    ...

    c6af84f

    fix(pr-pipeline): name each chunked finding's chunk and state the mul…

    ...

    d40517b
12. kyle-sexton commented on Oct 30, 2026

    

    kyle-sexton

    on Oct 30, 2026

    MemberAuthor

    More actions

    #11160 delivers part of this issue and now refers to it (Refs: #11000) instead of closing it.

    What #11160 delivers

    • pr-review-main reviews a diff over its 300000-byte cap in chunks: up to 4 chunks of at most 300000 bytes each, cut at file boundaries, with each file reviewed whole in one chunk.
    • Each chunk gets its own review run. The lane's one verdict sums the chunks' findings and sets critical to the most CRITICAL findings any one chunk returned (the max), and binds as a single-pass verdict does.
    • A chunk that fails, leaves no result or skips a part writes no verdict; the leg records verdict=none reason=chunk-failed and binds nothing.

    What remains open here

    • A multi-file diff over MAX_CHUNKS × 300000 bytes (about 1.2 MB) still stops as too-large. This issue asks for the size stop to fire only for a single file over the cap. Any fixed slot count leaves some multi-file diff over the cap, so lifting this needs a dynamic chunk count (for example, a planning job that emits a chunk matrix), which feat(pr-pipeline): review an oversize main-run diff in chunks #11160 does not build.