FREE TOOL / NO ACCOUNT / RUNS ON THE PAGE YOU GIVE IT
Is your site ready for AI agents?
Enter a URL and we request the same files an AI agent would: robots.txt, your sitemap, Link headers, llms.txt, the .well-known paths, and the homepage itself. You get a score out of 100 and the reason behind every point.
Free, no account, no card. We request the same files an AI agent would.
We ran this check across the startups paying to rank on outbid.lol and published every result with the fix. See how that board scored.
What this checks, and why each one matters
- AI crawler access
- Your robots.txt decides whether GPTBot, ClaudeBot, PerplexityBot and Google-Extended may read you at all. A site blocked here cannot appear in those models' answers, however good its content is.
- Discovery
- A valid sitemap, a Sitemap line in robots.txt, and RFC 8288 Link headers on the homepage. These are how an agent enumerates your pages instead of hoping every page is linked from the navigation.
- Agent interfaces
- /llms.txt, /.well-known/api-catalog, /.well-known/mcp/server-card.json, and whether asking for Accept: text/markdown returns markdown. These are new and rare. They are the difference between an agent reading your site and an agent using it.
- Agent-usable responses
- What your site says when the answer is no. An unknown path under /api/ should return a real status with a JSON body, not an HTML page a caller cannot parse, and a missing page should return 404 or 410 rather than 200 with an empty app shell.
- Machine readability
- A title, a meta description, valid JSON-LD, exactly one H1, whether the page still reads with JavaScript switched off, and how much of the HTML you send is readable text rather than markup. This is the part every model already relies on, and the part most sites can fix today.
- Identity and trust
- An about, contact and privacy page with real content, and an Organization schema that carries a contact point and an address. This is what an agent checks before it recommends a company it has not seen before.
QUESTIONS ABOUT THIS TOOL
- What is AI agent readiness?
- It is whether an AI agent can reach your site, find your pages, understand what they say, and get a usable answer when it asks for something. That breaks into six things: your robots.txt rules for AI crawlers, whether your pages are discoverable through a sitemap and Link headers, whether you publish the newer agent-facing files such as llms.txt or an MCP server card, whether your error responses are machine-readable, whether the page reads without JavaScript, and whether an agent can tell who you are and how to reach you.
- Which AI crawlers does this check?
- GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot (Perplexity) and Google-Extended (Google AI). For each one we read your robots.txt the way that crawler would and report whether it is allowed to read your site root.
- Does this cost anything or need an account?
- No. The check is our own HTTP requests to your public URLs and our own parsing of what comes back. No third-party data provider is involved, so there is nothing to bill and nothing to sign up for. It is rate limited to 10 checks an hour per address.
- What is llms.txt?
- An emerging convention: a markdown file at /llms.txt that lists the pages of your site that matter, with a short description of each, so a language model reads a curated map instead of guessing from your navigation. It is not an official standard and no model is required to honour it.
- Should I block AI crawlers?
- That is a business decision, not a technical one. Blocking them keeps your content out of those models, and it also removes you from the answers they give. This tool reports what your robots.txt currently says so the decision is at least a deliberate one.
- Why does the score feel harsh?
- Because a soft score is useless. We added checks for the things an agent has to do rather than only read: whether an unknown /api/ path answers JSON or an HTML page, whether a missing page returns a real 404, how much of your HTML is actually readable text, and whether your about, contact and privacy pages have anything in them. Our own site fails the JSON errors check, and we left it failing rather than dropping the check.
- What is the partial state for?
- Some things are half done rather than missing. A 404 that returns the right status but never says so in the body, an Organization schema with a name and no contact point, two of three trust pages: each of those is worth half a pass, so the score moves when you improve them instead of jumping only at the finish line.
- Why did a check fail on a file I do not have?
- Most sites do not publish an api-catalog or an MCP server card yet, so those checks fail for almost everyone today. They are in the score because they are where agent-facing discovery is heading, not because a missing file is currently costing you traffic.
A TOOL CHECKS ONCE. SEOBUILDER KEEPS WATCH.
We can watch this for you.
Your robots.txt, sitemap and structured data change every time someone ships. We run this check on a schedule, tell you when something that used to pass starts failing, and measure the part this page cannot: whether AI answers name you for your category, and who they name instead.