AI Contribution Policy Map

How … popular GitHub repositories treat AI-assisted contributions and AI agents that open pull requests. Each verdict links to the line it is based on.

Policies change. Verdicts come from keyword rules plus a review pass, and can be wrong or out of date. Read the linked source before you contribute.

Search repositories AI policy Agent PRs Language Sort

Other filters CLA DCO Issue first Outside PRs limited Anti-automation signals PR checkbox on AI use Reviewed only

JSON CSV

Repository	AI-assisted contributions	Agents opening PRs	CLA / DCO	Issue first	Outside PRs	Anti-automation

No repositories match these filters.

How this works

A collector reads each repository's CONTRIBUTING, AI policy, AGENTS.md / CLAUDE.md, pull request templates, workflows and labels (plus the organisation's .github fallback) through the GitHub API, and applies keyword rules. Each verdict keeps the file and line it came from. Where the rules disagree or only partly match, the verdict is Needs review instead of a guess. Verdicts marked reviewed were set by hand after reading the cited text; data/reviews.json records who reviewed each one and when. A review is flagged source changed when the cited file changes.

PR checkboxes are items in a repository's pull request template that ask about AI use or personal review, such as "I did not use AI to generate this PR" or "I am a human, and not a bot". Anti-automation signals include workflows and labels aimed at automated PRs, and lines in the docs that ask AI agents to identify themselves (for example with a marker in the PR title).

Not stated means the files we read say nothing we recognise. It does not mean AI contributions are welcome. Policies can also live in places we do not read (wikis, websites, discussions).

Maintainers: if a verdict is wrong or out of date, request a correction.

Data refreshed weekly. Last update: …. Code: MIT. Data: CC BY 4.0. Source on GitHub.