AGENTS.md

Think of it as a README written for agents: setup commands, how to run tests, code style, PR rules. It is stewarded by the Agentic AI Foundation under the Linux Foundation. The agents.md site lists Codex, Cursor, Gemini CLI, Jules, VS Code, Zed, Devin, Windsurf and others as supporting it.

When to use it

• More than one coding agent touches your repo and you want one set of instructions.
• You want tests and lint to run before an agent calls a task done. List the commands and agents will try them.
• A monorepo: put an AGENTS.md in each package. The closest one to the edited file wins.

How the big three read it (from their docs, 2026-09-25)

• Codex reads ~/.codex/AGENTS.md first, then every AGENTS.md from the Git root down to the current directory, joined in order. AGENTS.override.md takes priority in a folder. The combined text stops at 32 KiB by default (project_doc_max_bytes).
• Claude Code (v2.1.277+) reads AGENTS.md only when there’s no CLAUDE.md or CLAUDE.local.md in the working directory or above. If you have both, put @AGENTS.md in your CLAUDE.md to import it.
• Gemini CLI, per the agents.md FAQ, needs "context": {"fileName": "AGENTS.md"} in its settings.

Before you rely on it

Agents run the commands you list, so treat AGENTS.md like a script: review changes to it in pull requests, especially in repos that take outside contributions.