LearnNewsExamplesServices
Frontmatter
id13727
titlefeat(ai): PR-outcome revert-detection core (pure) — slice 2 of #9962
stateClosed
labels
aiarchitecturemodel-experience
assignees[]
createdAtJun 21, 2026, 10:03 AM
updatedAtJun 21, 2026, 11:48 AM
githubUrlhttps://github.com/neomjs/neo/issues/13727
authorneo-opus-ada
commentsCount0
parentIssuenull
subIssues[]
subIssuesCompleted0
subIssuesTotal0
contentTrust
projected
quarantined0
signals[]
blockedBy[]
blocking[]
closedAtJun 21, 2026, 11:48 AM

feat(ai): PR-outcome revert-detection core (pure) — slice 2 of #9962

Closed v13.1.0/archive-v13-1-0-chunk-5 aiarchitecturemodel-experience
neo-opus-ada
neo-opus-ada commented on Jun 21, 2026, 10:03 AM

Slice 2 of #9962 (PR Outcome Tracker — RLAIF reward signal), sibling of #13724.

Scope

The PURE revert-commit-trailer parsing core (no I/O), producing the reverted flag that #13724's classifyPrOutcome consumes for the -1.0 "actively harmful — regression" reward — the highest-value, hardest reward signal (a session whose merged PRs were later reverted must score negative; @neo-opus-grace agreed in the #9962 design-response: revert via the trailer, not closedAt).

Deliverable: pure exports —

  • parseRevertTrailer(commitMessage) → the reverted SHA(s) from a This reverts commit <sha> trailer, or null.
  • isRevertOf(commitMessage, targetSha) → boolean (does this commit revert the target?).

Fully unit-tested (trailer present/absent, short vs full SHA, multiple reverts, malformed).

Out of scope (stays on #9962, @neo-opus-grace's integration)

The commit SCAN (gh/git I/O to fetch later commits), wiring the detection into the outcome-scan, and feeding the result to classifyPrOutcome. This slice is the pure parse only — the same carve as #13724.

Refs #9962, #13724.

tobiu referenced in commit efc47d1 - "feat(ai): pure PR-outcome revert-detection core (#13727) (#13729) on Jun 21, 2026, 11:48 AM
tobiu closed this issue on Jun 21, 2026, 11:48 AM