AI-assisted contributions are allowed.
Unreviewed AI-generated contributions are not.
If you submit an issue, pull request, comment, review, documentation change, or code contribution, you are responsible for its content.
Before submitting AI-assisted content, you must:
- read it;
- understand it;
- verify it;
- test it when applicable;
- make sure it actually reflects what you want to contribute.
Do not submit raw AI output and expect maintainers to review it for you.
If AI helped generate code, you are still expected to understand the resulting changes.
You should be able to explain:
- what the code does;
- why the change is necessary;
- why the chosen implementation makes sense;
- what you tested.
If you cannot explain your own pull request without asking an AI tool, it is not ready to be submitted.
AI may help you structure or improve an issue, but the final content must describe a real problem you understand.
Do not submit speculative AI-generated bug reports, root-cause analyses, feature requests, or large explanations that you have not verified yourself.
A short and accurate issue is more useful than a long generated one.
Do not blindly paste maintainer feedback into an AI tool and paste the response back.
AI can help you investigate or formulate an answer, but you are expected to read, understand, and agree with anything you post.
AI is a "tool", not a contributor.
Do not blame it for incorrect code, bad documentation, invented APIs, security issues, or misleading information.
If you submit it, you own it.
Maintainers may close or reject contributions that appear to be:
- blindly AI-generated;
- unreviewed;
- unnecessarily verbose;
- unrelated to the reported problem;
- technically incorrect;
- submitted by someone who does not understand the contribution.
The goal of this policy is not to discourage the use of AI.
The goal is simple:
Use AI to help you do the work. Do not make maintainers review AI output that you have not reviewed yourself.