How agent-ready are your docs?
Docs for Agents
Your documentation has AI readers now.
Docs for Agents audits what an AI agent receives from each product’s public documentation, then publishes the score and the evidence for every check.
Every report ends in a scorecard.
These cards show the top score in each letter-grade band, from A to F. The grades table ranks all 97 products.
9 of 10 checks earn full points. Largest gap: Agent interoperability signals.
6 of 10 checks earn full points. Largest gap: API description.
6 of 10 checks earn full points. Largest gap: Markdown mirror.
5 of 10 checks earn full points. Largest gap: Markdown mirror.
5 of 10 checks earn full points. Largest gap: llms.txt.
Ten checks, each with the response as evidence.
The audit requests the docs the way an agent does: the page without JavaScript, a Markdown version, llms.txt, robots.txt, the sitemap, and Model Context Protocol endpoints. Each check earns points out of 100 and links to the exact URL it requested. No AI model scores anything.
How the test worksMost docs reach agents. Fewer are built for them.
93 of 97 products put their page content in the HTML response, and 93 publish a valid llms.txt. Only 60 keep llms.txt small enough for one agent request. 75 serve a Markdown version that matches the page, and 46 run an MCP server that answers without sign-in.
How agent-ready are your docs?
Agent-ready documentation reaches an AI agent in a form it can use: readable without a browser, available as Markdown, indexed in llms.txt, open to AI crawlers, and reachable through tools such as an MCP server. The test measures each of these directly and publishes the evidence. The scorecard shows where an agent gets what it needs and where it does not.
Frequently asked questions
Does an AI model grade the docs?
No. Every check is a direct HTTP request to the public docs, scored by fixed rules. Anyone can repeat a request and see the same response.
What is the AI Agent Readiness Score?
A score out of 100 points: content delivery is worth 50, discovery 25, and agent capabilities 25. The percentage becomes a US school letter grade. The docs platform is recorded but never scored.
What happened to the reading test?
From August to October 2026, three AI models read each product’s docs and attempted five first-hour developer jobs. That reading test was retired on October 16, 2026. Its reports stay in the reading test archive with their original scores.
Can a score change?
Yes. A score reflects the docs on its audit date. A new audit replaces the score and prints its own date.
Can you test our docs?
Nominate the docs site on the nominate page. The audit uses public pages only.
The next report needs a nominee.
Nominate a docs site