bench(reviewer-eval): waive hunk drops for rebutted candidates — decoys don't need the hunk

[norvalbv]

What

propose.mts's empty-hunk / hunk-too-long hard drops no longer apply to outcome === 'rebutted' candidates. Golds (fixed) keep both checks unchanged.

Why

Rebutted bot threads are the scarce mined-decoy source. A decoy fixture is built from baseFileContent (contents API pinned to originalCommitId) plus the bot's rebutted claim — the diff hunk's job is to locate a defect being fixed, which a decoy by definition doesn't have. Dropping a rebutted candidate because its hunk was truncated (mine-bots caps at 4000 chars) throws away exactly the rows the corpus needs most (55 decoys of the ~120-row revisit threshold).

Numbers (re-derived from candidates.jsonl, not the handover's estimate)

The handover claimed rebutted threads "mostly fail hard drops" — that was wrong. Of 749 candidates, 34 are rebutted: 25 already survived, 4 died hunk-too-long, 1 empty-hunk, 4 already-in-corpus. The 12 rebutted rows dropped as out-of-charter are genuinely out of charter (Maintainability/docs comments on baselines and decision docs) — the router is right and stays untouched. no-line also stays: zero rebutted rows die there, and a row with neither line nor hunk gives the adapter nothing to anchor the claim to.

Verified live against the real pool: correctness routing 173 → 176 (+3 in-charter decoy candidates); hunk-too-long 113 → 109; empty-hunk drop gone; all other counts identical. The priority sort already ranks rebutted after fixed, so freed decoys queue behind golds.

🤖 Generated with Claude Code

[coderabbitai]

Warning

Review limit reached

@norvalbv, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 4 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 08d92115-0dc4-4dcf-9da9-71604432aae4

📥 Commits

Reviewing files that changed from the base of the PR and between 41348e6 and edf66f4.

📒 Files selected for processing (1)

• gate-engine/review/eval/reviewers/propose.mts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}