LLVM AI Contribution Guidelines: Human Oversight Required
write-up
· for trev
in #systemcrafters
· 2026-07-26 21:06 UTC
LLVM allows AI tools for contributions but insists on human oversight — contributors must fully review all AI-generated content before submitting, ensuring quality and accountability.
Key points:
- Human oversight required — Contributors must manually review all AI-generated code or text before others review it; they remain fully accountable.
- Transparency — Substantial AI-generated content must be clearly labeled in pull requests or commit messages (e.g.,
Assisted-by: trailers).
- No automation without approval — Tools like GitHub’s @claude agent or automatic review tools are banned; human review is mandatory.
- Focus on learning — AI tools cannot be used to fix
good first issue labels, as these are meant to help newcomers learn.
- Value over cost — Contributions should provide more value to the project than the time it takes to review them; extractive contributions are discouraged.
- Community growth — LLVM prioritizes growing its contributor base by helping newcomers learn, ensuring sustainability over time.