fix(benchmarks): point SWE-bench metadata at README

[lalalune]

Summary

• fixes the SWE-bench Python package metadata so project.readme points at the existing README.md instead of deleted RESEARCH.md
• adds issue evidence for the metadata/package build validation

Validation

• python3 - <<'PY' ... PY metadata check: README.md exists
• python3 -m pytest packages/benchmarks/swe_bench/tests -q (131 passed)
• python3 -m build packages/benchmarks/swe_bench --sdist --wheel --outdir /tmp/eliza-swebench-build
• git diff --check

Evidence

• .github/issue-evidence/11346-swebench-readme-metadata/README.md

Issue

Progress toward #11346. This does not close the full benchmark-matrix issue; it removes a packaging/readiness break in the SWE-bench harness metadata.

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on this repository. 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: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: a26a84e7-93e9-4090-8c58-3f12d5d65678

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 fix/11346-swebench-readme-metadata

---

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.}

[greptile-apps]

Your trial has ended. Reactivate Greptile to resume code reviews.

[lalalune]

Same-account review note, not an approval.

I found one small packaging-doc follow-up and pushed it to this PR branch:

• be174c270e docs(swe-bench): remove stale research readme reference
• Reason: after project.readme was pointed at README.md, that README still told package readers to see deleted RESEARCH.md. The follow-up makes the README self-contained and records the Windows validation in the evidence file.

Local Windows validation after the follow-up:

• Parsed packages/benchmarks/swe_bench/pyproject.toml; project.readme == "README.md" and the file exists.
• rg -n "RESEARCH\.md" packages/benchmarks/swe_bench -> no matches.
• python -m build packages/benchmarks/swe_bench --sdist --wheel --outdir %TEMP%/eliza-swebench-build-11620 -> built elizaos_swe_bench-2.0.0.tar.gz and elizaos_swe_bench-2.0.0-py3-none-any.whl successfully.
• git diff --check -> clean.
• python -m pytest packages/benchmarks/swe_bench/tests -q reached 124/131 locally; failures were environment-specific and unrelated to this metadata/doc change: two Windows fake bash/opencode executable tests (WinError 193) plus five tests requiring the missing datasets Python package.

No benchmark execution/scoring/model behavior changed, so real-model trajectory and screenshot/video evidence remain N/A for this PR.

[github-actions]

LifeOps Multi-Tier Benchmark

Suite: — Tiers requested:

large — cancelled

frontier — cancelled

Artifacts: lifeops-multi-tier-large-28624463888, lifeops-multi-tier-frontier-28624463888

[claude]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.