AI Agent Readiness Checker
Test any website for the signals AI agents look for: llms.txt, AI-crawler permissions, an OpenAPI spec, developer docs, and MCP support. Get a 0–100 readiness score.
Why agent readiness matters
AI assistants and autonomous agents increasingly decide which products get recommended and which APIs get called. A site that publishes llms.txt, allows AI crawlers in robots.txt, exposes an OpenAPI spec, and documents an MCP server is dramatically easier for agents to discover, understand, and integrate. This checker probes those exact surfaces — the same signals behind our directory-wide Agent Readiness leaderboard.
What the checker actually probes
Submitting a URL fires seven parallel requests at the domain root: /llms.txt, /robots.txt, /openapi.json, /.well-known/openapi.json, /.well-known/ai-plugin.json, /docs, and the homepage itself. Responses that come back as an HTML page are discarded, because single-page apps routinely return their app shell for any unknown path and a wildcard 200 is not a real file. The homepage HTML is then scanned for references to a public API and to MCP. Each passing check adds its fixed weight, the total is graded, and every individual check is listed with its points so you can see exactly where the score came from.
The tool works on any public site, not only companies listed in the directory, and it is aimed at product, docs, and developer-relations teams who want an outside view of how legible their site is to an assistant deciding what to recommend or call. Because every probe is an unauthenticated GET with a short timeout, sites behind login walls, aggressive bot protection, or a strict WAF can score lower than their real capabilities justify — treat a low score as a prompt to check those paths manually rather than a verdict.
Frequently asked questions
›What is an AI agent-readiness score?
It is a 0–100 measure of how easily an AI assistant or autonomous agent can discover, understand, and act on your site. The score rewards machine-readable surfaces — an llms.txt file, an OpenAPI spec, reachable developer docs, an MCP endpoint, and crawler rules that let AI bots in. A high score does not mean your product is good; it means an agent can find out what your product does without a human in the loop.
›How is the readiness score calculated?
Seven checks run against the domain and each contributes fixed points: llms.txt present (20), robots.txt does not block AI crawlers (15), a discoverable OpenAPI spec (15), reachable developer docs at /docs (15), a public API referenced on the homepage (15), MCP mentioned on the homepage (15), and a .well-known/ai-plugin.json manifest (5). Anything at 70 or above is labelled agent-ready, 40–69 partially ready, and below 40 not agent-ready yet.
›What is llms.txt and does my site need one?
llms.txt is a plain-text file at the root of a domain that points language models to the pages worth reading — docs, API references, pricing — in a compact, link-first format. It is a community convention rather than an official standard, and no model is obliged to read it. It is still the single highest-weighted check here because publishing one is cheap and it tells an agent where to look instead of making it guess.
›Does blocking GPTBot or ClaudeBot lower my score?
Yes, if your robots.txt disallows the whole site for the AI crawlers it names. The check looks for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot, and Anthropic rules and only fails when every one of them is blocked site-wide. Blocking crawlers can be a legitimate business decision — this tool simply reports the trade-off, since a blocked site cannot be cited or recommended by the assistants using those crawlers.
›My site scored low. What should I fix first?
Start with the highest-weighted gaps you can close quickly: publish an llms.txt, make sure /docs resolves, and confirm robots.txt is not blocking AI crawlers wholesale. If you have an API, expose the OpenAPI spec at a predictable path and link the developer docs from your homepage. Shipping an MCP server is the bigger lift but the one that most directly lets agents use your product rather than just read about it.
›Is the agent readiness checker free?
Yes — free, no signup, and it works on any public website, not only startups listed in NeuronFeed. Checks are rate-limited to 20 per minute per IP address, and results are returned live rather than stored.