feat(competition): relevance-guard engine-lane submission

[plind-junior]

engine-lane entry for season 1 (#573): relevance-guarded ranking.

one general idea carried through: every class of suspect memory — hearsay, third-person possessive fragments, stored instructions, stale values contradicted by a change-of-state memory — gets a single uniform penalty scaled by how much the suspect matches the query. suspects fall out of a tight candidate pool in damage order, so the item the question is actually about is the one excluded first. past-tense questions ("what was ... before it changed?") suspend the collapse, because there the stale side is the answer. no benchmark-keyed logic: no value lists, no template matching, no category dispatch.

local verification with the ci-exact loop:

vouch bench run --seeds 1,2,3,4,5,6,7,8,9,10,11,12 \
    --strategy contrib/strategies/relevance_guard.py \
    --against contrib/strategies/baseline.py
challenger 0.7083  champion 0.6583  diff +0.0500  band 0.0381
DETHRONED

per the lane rules the gate scores this in a sandbox and it never auto-merges; on a human merge above the band, the ledger sweep promotes it to reigning champion and the threshold ratchets.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

🗂️ Base branches to auto review (2)

• test
• main

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 660925d3-0ce4-4537-b0fd-784e69406c91

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch engine/relevance-guard

---

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.}

[github-actions]

koth engine lane - dethroned

{
 "date": "2026-07-28",
 "base_sha": "b11ac52a1fbd1172887eb924a8e170aec15eef5a",
 "seeds": [
  67078370866574,
  116114090607205,
  108999320470477,
  114845009661335,
  248394666013973,
  108843862080367,
  274140852163751,
  57307557147568
 ],
 "lane": "engine",
 "champion": {
  "scores": [
   0.5,
   0.7,
   0.5,
   0.7,
   0.7,
   0.8,
   0.7,
   0.7
  ],
  "mean": 0.6625
 },
 "challenger": {
  "scores": [
   0.5,
   0.7,
   0.6,
   0.7,
   0.7,
   0.9,
   0.8,
   0.7
  ],
  "mean": 0.7
 },
 "mean_diff": 0.037500000000000006,
 "se": 0.018298126367784998,
 "band": 0.03586432768085859,
 "dethroned": true
}

engine code is NOT auto-merged. a winning strategy earns the
leaderboard place; a maintainer reviews the code and merges it
as a new default. the daily result is provisional (public seeds)

• payout rank is settled by the monthly sealed run.

[github-actions]

koth engine lane - dethroned

{
 "date": "2026-07-28",
 "base_sha": "b11ac52a1fbd1172887eb924a8e170aec15eef5a",
 "seeds": [
  67078370866574,
  116114090607205,
  108999320470477,
  114845009661335,
  248394666013973,
  108843862080367,
  274140852163751,
  57307557147568
 ],
 "lane": "engine",
 "champion": {
  "scores": [
   0.5,
   0.7,
   0.5,
   0.7,
   0.7,
   0.8,
   0.7,
   0.7
  ],
  "mean": 0.6625
 },
 "challenger": {
  "scores": [
   0.5,
   0.7,
   0.6,
   0.7,
   0.7,
   0.9,
   0.8,
   0.7
  ],
  "mean": 0.7
 },
 "mean_diff": 0.037500000000000006,
 "se": 0.018298126367784998,
 "band": 0.03586432768085859,
 "dethroned": true
}

engine code is NOT auto-merged. a winning strategy earns the
leaderboard place; a maintainer reviews the code and merges it
as a new default. the daily result is provisional (public seeds)

• payout rank is settled by the monthly sealed run.