About Agent-Ready Check

What this is

A single-page scanner for the standards that make a website usable by AI agents. It fetches the well-known documents, headers and DNS records an agent would look for, reports each one as pass, warning, issue or note, and maps the result onto a five-level readiness ladder.

Why evidence instead of a score alone

A number you cannot audit is not useful. Every check in the results list expands to show the exact request that was made, the response that came back and the finding it produced, so you can disagree with the scanner and be right.

What it deliberately does not do

It does not tell you to publish documents for infrastructure you do not run. Five commerce checks and one informational check are excluded from the score when they do not apply, because sending agents to a 404 is worse than admitting you have no store.

This site scans itself

The site you are reading publishes its own robots.txt, sitemap, llms.txt, API catalog, MCP Server Card, A2A Agent Card, Agent Skills index, ARD manifest and auth.md, and serves markdown to agents that ask for it. Run the scanner against this domain to see the result.

Open source

The check registry, the level ladder and the fix prompts are all in one repository. Add a locale, change the ladder, or delete the checks you disagree with.

Use it from an agent

Three machine interfaces, all on this domain:

• POST /api/scan: JSON in, JSON out.
• POST /api/scan with "format": "agent" returns the same result as markdown sized for a model.
• POST /mcp: a Model Context Protocol server exposing a scan_site tool.