AGENTS.md guide
How to write AGENTS.md so coding agents respect repo boundaries, tests, and review norms.
AGENTS.md is a repo-level instruction file aimed at autonomous coding agents—complementing human README docs with enforceable habits.
What to include
- Setup — install, env vars (names only), default
SITE_IDor app entrypoints - Verification — exact test/lint commands agents must run before claiming done
- Scope rules — directories off-limits, no force-push, no secret files
- Commit norms — when to ask the user before committing
- Architecture — one paragraph on how packages relate
What to avoid
- Paste credentials or production URLs with tokens
- Vague “be careful” without naming commands
- Duplicating entire style guides—link to linters instead
This monorepo’s apps/ai-network/AGENTS.md is a living example—trimmed per app as the network grows.
Pair with Cursor rules guide when teammates split between IDE and CLI agents.