AGENTS.md

Detects an AGENTS.md file: a plain-Markdown instructions file aimed at coding agents (Claude Code, Cursor, Codex, and similar) that work against a repository or a documentation site. Where llms.txt curates content for an agent reading a site, AGENTS.md tells an agent working on the underlying project how to build, test, and contribute to it.

What it checks

The audit probes, once per crawl, for AGENTS.md at a few conventional locations, in order:

• /AGENTS.md (root, canonical casing)
• /agents.md (lowercase)
• /.well-known/agents.md
• /docs/AGENTS.md

It reports which path (if any) responds, and does a basic content sanity check: the response should be plain text or Markdown, non-empty, and not an HTML error or SPA-fallback page mistakenly served with a 200.

Solution

If the site has an associated codebase, publish an AGENTS.md at the repository root (it will typically be served at /AGENTS.md if the site is generated from that repo, or link to it from your docs). Cover what a coding agent needs to be productive without asking:

# Agent Instructions

## Setup
Run `bun install` then `bun run dev`.

## Testing
Run `bun test` before committing. All tests must pass.

## Conventions
- TypeScript strict mode, no `any`.
- Components in `src/components`, one per file.

## Do not
- Do not edit generated files in `src/generated/`.
- Do not commit directly to `main`.

For a documentation site without its own codebase, AGENTS.md is less relevant — prioritize llms.txt instead.

Enable / Disable

Disable this rule

[rules]
disable = ["ax/agents-md"]

Disable all Agent Experience rules

[rules]
disable = ["ax/*"]

Enable only this rule

[rules]
enable = ["ax/agents-md"]
disable = ["*"]