AGENTS.md
AGENTS.md is an open convention for a markdown file, placed at the root of a code repository, that briefs AI coding agents on how to work with the project: setup commands, build and test steps, code conventions, and constraints. It works as a README for agents — a dedicated, predictable place for the detailed context coding agents need but that would clutter a human-facing README.
The format is deliberately unowned: plain markdown, no required schema, one file that a broad ecosystem of coding agents reads — including OpenAI's Codex, Google's Jules and Gemini CLI, Cursor, Devin, and GitHub Copilot's coding agent, per the agents.md compatibility list. Adoption spans tens of thousands of open-source repositories. Because the file lives in the repo, it travels with the code: any agent that clones or browses the project finds its operating instructions in the same place every time.
For an API producer, the relevant repository is usually the primary SDK or the flagship example repo. Coding agents are the first mass agent-readers of APIs — every scaffolded integration is an agent reading your docs — and an AGENTS.md there can state the things reference docs rarely front-load: which auth pattern the SDK expects, how to run the test suite against a sandbox, which endpoints are deprecated. A stub file that says nothing an agent can act on delivers the signal without the substance.
Discry's AGENTS.md discovery check looks for the file in the API's primary public GitHub repository on its default branch (with the org-level .github repository as a fallback). A file with substantive content passes; a present-but-minimal stub earns partial credit; no file, or no known repository, fails.
Checks: AGENTS.md · how each is scored
What does an AI agent make of your API?
Find out in about a minute — no signup.
Discry your API — free