GitHub frames mentorship around 3 Cs

The sharp part here is that GitHub is not treating AI-assisted contributions as a quality problem so much as a trust-signal problem. The framework is simple enough to adopt immediately, and that is probably why it works.

• –`Comprehension`: ask for issue-first discussion or live conversation before code lands, so you can tell whether the contributor understands the problem.
• –`Context`: require enough detail to review well, including issue links, tradeoffs, and AI disclosure where relevant.
• –`Continuity`: spend real mentorship time only after someone comes back, responds thoughtfully, and shows they want to grow in the project.
• –Practical upside: it gives maintainers a rubric instead of vibes, which should reduce burnout and bias.
• –Practical downside: projects will need to enforce these filters consistently, or they become another soft guideline that AI can still route around.