chore(master): release 2.3.1

[github-actions]

🤖 I have created a release beep boop

2.3.1 (2026-05-11)

Bug Fixes

• eval: align ContextBench harness evidence contracts (4513979)
• eval: deduplicate blocked ContextBench rows (99c9753)
• eval: deduplicate blocked ContextBench rows (c41e844)
• eval: harden ContextBench fixture verification (bed5064)
• eval: harden ContextBench manifest checks (04a6cfb)
• eval: preserve ContextBench executor model provenance (867ac70)
• format: format ContextBench harness sources (b2fa208)
• git: scope local artifact ignores (ef42e53)
• test: harden ContextBench schema cleanup (c5a74af)
• test: isolate ContextBench baseline Git env (6aed9d1)
• test: isolate ContextBench git fixtures (62d3110)
• test: relax slow Windows integration timeouts (5675ebd)
• test: relax slow Windows search timeouts (cad646d)
• test: relax zombie guard timeout jitter (5a5bf68)
• test: tolerate ContextBench runner cleanup races (c027703)
• test: tolerate ContextBench schema cleanup races (a155d56)
• test: tolerate ContextBench temp cleanup races (0360cb9)

---

This PR was generated with Release Please. See documentation.

[greptile-apps]

Greptile Summary

Automated patch release PR generated by Release Please, bumping the package from 2.3.0 to 2.3.1 across package.json, .release-please-manifest.json, and CHANGELOG.md with 17 bug-fix entries covering ContextBench eval hardening, test isolation, and timeout relaxation.

• The CHANGELOG.md contains two entries for eval: deduplicate blocked ContextBench rows from commits 99c9753 and c41e844 — worth confirming these are genuinely distinct commits and not an accidental double-apply or cherry-pick.
• The repo's release checklist (AGENTS.md) requires README.md and docs/capabilities.md to be updated alongside any version bump; neither file appears in this PR.

Confidence Score: 4/5

Safe to merge; the only substantive changes are version numbers and a CHANGELOG prepend generated by automation.

The three changed files are all release metadata — no runtime logic is touched. The duplicate CHANGELOG entry is cosmetic but worth a quick check to ensure the underlying git history doesn't contain a redundant commit. The release checklist gap (missing README/docs updates) may or may not matter depending on whether the fixed bugs affect documented capabilities.

CHANGELOG.md — duplicate fix entry and missing companion doc updates per the release checklist.

Important Files Changed

Filename	Overview
.release-please-manifest.json	Version bumped from 2.3.0 to 2.3.1 by Release Please automation — no issues.
package.json	Package version bumped from 2.3.0 to 2.3.1 — standard automated patch release change.
CHANGELOG.md	2.3.1 section added; contains a duplicate entry for "eval: deduplicate blocked ContextBench rows" from two commits with identical messages, suggesting the same fix may have been applied twice in the git history.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Release Please Bot] --> B[Bump .release-please-manifest.json\n2.3.0 → 2.3.1]
    A --> C[Bump package.json\nversion 2.3.0 → 2.3.1]
    A --> D[Prepend CHANGELOG.md\nwith 2.3.1 section]
    D --> E{Duplicate entry?\neval: deduplicate blocked\nContextBench rows}
    E -->|Yes - 2 commits,\nsame message| F[⚠️ Verify commits\n99c9753 & c41e844\nare distinct]
    E -->|No issue| G[✅ CHANGELOG OK]
    B --> H[Merge to master]
    C --> H
    G --> H
    F --> H

Loading

_{Reviews (1): Last reviewed commit: "chore(master): release 2.3.1" | Re-trigger Greptile}

[greptile-apps]

Duplicate fix entry with identical commit message — eval: deduplicate blocked ContextBench rows appears twice (commits 99c9753 and c41e844). Two separate commits carrying the exact same message suggests the patch was cherry-picked or applied a second time unintentionally. Worth confirming that both commits represent genuinely distinct changes; if one is a duplicate commit in the git history, the extra commit should be dropped before merging the release.

[greptile-apps]

Release checklist not fully satisfied — AGENTS.md requires that README.md and docs/capabilities.md are updated before any version bump. This PR only touches CHANGELOG.md, package.json, and .release-please-manifest.json. If those documents contain capability or usage information that reflects the 2.3.1 bug fixes, they should be updated (or explicitly confirmed as needing no change) before the release is merged.

Context Used: AGENTS.md (source)