AGENTS.md vs CLAUDE.md vs llms.txt

AGENTS.md is an open convention: a plain-Markdown file at your repo root that tells coding agents what they need to know to work in your project - build and test commands, code style, conventions, and gotchas that would clutter a human README. There's no required schema; it's just standard Markdown.

It started across tools like OpenAI Codex, Cursor, and Jules, and is now stewarded by the Agentic AI Foundation under the Linux Foundation. A wide range of agents read it, which is the point: write it once, and many tools use it.

CLAUDE.md is Claude Code's native memory file. It's loaded at the start of every session and holds the same kind of context - commands, conventions, architecture, standing instructions - but it's specific to Claude Code, with support for scoped files (org, user, project, local) and @-imports.

Since v2.1.277, Claude Code reads AGENTS.md as project instructions when the repo has no CLAUDE.md or CLAUDE.local.md; if a CLAUDE.md exists, Claude Code reads that instead by default. The recommended pattern when you keep both is a CLAUDE.md that imports AGENTS.md (via an @AGENTS.md import or a symlink), so both tools share one source of truth and you add Claude-specific notes below.

llms.txt is a proposed Markdown index for a website. An H1 names the project; an optional summary and links help an agent find relevant documentation. The August 2026 v2 proposal allows /llms.txt at the root or a scoped path such as /docs/llms.txt, and recommends linking to Markdown versions of pages.

Some sites also publish llms-full.txt with concatenated content. A concise index lets an agent follow the links it needs. Publishing either file does not guarantee that a particular tool will read it or that your pages will appear in AI search results.

Checked September 16, 2026. Native discovery and an explicit import are different behaviors; a file name alone does not make every tool read it.

Sources: Codex instruction discovery, Cursor rules, Claude Code memory and imports, and the llms.txt proposal.

For a web project, the source directory serving public files may differ by framework. The important distinction is that the first two files live in the repository, while llms.txt is served by the website at its root or a scoped path.

my-project/
  AGENTS.md
  CLAUDE.md
  services/
    billing/
      AGENTS.md
  public/
    llms.txt       -> https://example.com/llms.txt

Start with a small shared file. Replace the example commands with commands that actually exist in your repository.

# Project instructions

## Verification
- Run npm run lint and npm test after source changes.
- Run npm run build before handing back an app change.

## Conventions
- Reuse the shared UI components.
- Update the setup guide when required environment variables change.

Save that as AGENTS.md. For Claude Code, add a CLAUDE.md containing an explicit import and any Claude-specific notes:

@AGENTS.md

## Project context
Read docs/architecture.md before changing service boundaries.

Build your AGENTS.md or create a CLAUDE.md. For file precedence and maintenance, use the AGENTS.md guide and CLAUDE.md guide.

Use links to public documentation you want readers to find. Publishing this file is optional and does not replace your sitemap, HTML content, or access controls.

# Example API

> Documentation for the Example API.

## Guides
- [Quickstart](https://example.com/docs/quickstart): Send your first request.
- [Authentication](https://example.com/docs/auth): Configure an API token.

Generate an llms.txt, then validate its structure. Google says its AI search features do not require special optimization.