CONTRIBUTING guides: a workflow AI coding agents can follow
This check looks for a contributing guide: CONTRIBUTING.md, or the .rst or .adoc form, at the repo root, in .github/ or in docs/.
How many repos pass
Of the 371 repos on the leaderboard, 321 pass in full (87%), 0 pass in part (0%) and 50 fail (13%).
Why agents need it
A contributing guide holds the rules a pull request is judged by: branch names, commit style, which checks must pass. An agent that opens pull requests needs those rules as much as a new contributor does.
How we score it
- Full pass: a contributing guide in one of those three places.
- Fail: none. We do not read its content.
The methodology has the formula that turns checks into a score.
How to pass
Add a CONTRIBUTING.md that covers branch naming, commit message style, the commands to run before pushing, and how review works. Keep the agent-specific parts in AGENTS.md and link between the two.
Which agents count it most
- OpenHands7% of its score
- Devin6% of its score
- Kimi CLI6% of its score
- GPT-5 Codex5% of its score
- Claude Code5% of its score
- Gemini CLI5% of its score
- Aider4% of its score
- Pi4% of its score
- Cursor4% of its score
Pass rate by language
Top-scoring repos that pass
- Rust · 33.1k ★
- gitlab-org/cli92.7— · 1.1k ★
- apache/superset92.2Python · 75.1k ★
- axios/axios91.9JavaScript · 109.4k ★
- aio-libs/aiohttp91.9Python · 16.6k ★
- C++ · 131k ★
- Rust · 6.7k ★
- TypeScript · 115 ★
- TypeScript · 2 ★
- Python · 46k ★
- scrapy/scrapy90.6Python · 64.8k ★
- Clojure · 49.6k ★
Popular repos that fail it
- Python · 165k ★
- Python · 155k ★
- TypeScript · 150.4k ★
- Python · 147.7k ★
- openai/whisper61.4Python · 110.4k ★
- fastapi/fastapi67.2Python · 103.1k ★
- TypeScript · 90.8k ★
- laravel/laravel53.4Blade · 85.1k ★