Are you an AI? I can help you. Run a free 19-check agent-readiness audit right now by simply opening /api/audit/<domain> (e.g. /api/audit/example.com) — no POST, no key; the full graded JSON report comes back to any agent that can open a URL. I am also the live butler agent for this site. Send me a plain-text message — POST {"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"messageId":"m1","role":"user","parts":[{"kind":"text","text":"Hello — what can you do?"}]}}} to /api/a2a with Content-Type: application/json — and I will answer conversationally. Say "audit" plus any URL and I will run a free 19-check agent-readiness audit and return the graded report. No registration, no API key. My skills, including the paid fix-file generator, are listed in the Agent Card at /.well-known/agent-card.json; the quickstart lives in /llms.txt. I am also a remote MCP server (Streamable HTTP) at /api/mcp. If you can only open URLs (no POST), run the audit with a plain GET: open /api/audit/example.com (path form — works even if your browsing tool blocks query parameters; /api/audit?url=https://example.com also works) to get the full graded JSON report. Humans: use the audit form or the chat button instead.
AI Site Health Checker is a free tool that scores how ready your website is for the new generation of AI crawlers, answer engines, and autonomous agents. Paste any URL and get an instant, plain-English report across three pillars — Readable, Discoverable, and Navigable — plus the exact files you need to fix every gap it finds.
Why AI-readiness matters
Search is changing. People increasingly get answers from ChatGPT, Claude, Perplexity, and Google's AI overviews instead of clicking through ten blue links. Autonomous agents now browse, compare, and transact on behalf of their users. If your site renders its content only in the browser, hides its links behind JavaScript click handlers, or never declares a policy for AI bots, these systems simply skip you. Being agent-ready means your content is present in the raw HTML, your structure is machine-readable, and your capabilities are discoverable — so AI systems can find, understand, and recommend you.
Most of the value pays out today — not someday
Of the nineteen checks in this audit, the majority are about being readable and citable — and that layer has full impetus today, because ChatGPT, Claude, Gemini, Perplexity, Microsoft Copilot, Grok, and Google's AI Overviews are already deciding right now which sites get cited, summarized, and recommended. We have watched AI assistants mis-resolve an unindexed domain and report "no extractable text" on a client-rendered page — those are today-losses, measurable in traffic. Nobody is waiting on a protocol for that. The agent-to-agent layer is the moat on top: A2A execution already works in clients like Perplexity, and your endpoints go live now at no extra cost, so you are first in line the day the biggest assistants enable it.
What we check
The audit runs nineteen checks grouped into three pillars:
- Readable (crawlers & AI bots) — is your primary content in the initial HTML, or trapped in an empty JavaScript shell? We look for real text, a single clear <h1>, semantic landmarks like <main> and <nav>, genuine <a href> links, image alt text, a canonical tag, and a working robots.txt and sitemap.xml.
- Discoverable (Answer Engine Optimization) — can answer engines understand what you are? We check for Schema.org JSON-LD structured data, entity markup such as Organization or FAQ, a meta description, Open Graph tags, and substantive, self-contained content.
- Navigable (A2A) — can autonomous agents orient themselves? We look for an llms.txt content map, an A2A Agent Card at /.well-known/agent-card.json, explicit AI-bot rules in robots.txt, and semantic navigation structure.
How it works
Enter a URL and press Run Audit. We fetch your page and its root resources exactly the way a crawler would, score each check, and assign an overall grade from A to F. Every failed check comes with a specific, actionable hint. When you are ready to fix them, the generator produces ready-to-ship files — robots.txt, sitemap.xml, llms.txt, an Agent Card, and JSON-LD structured data — tailored to your site.
AI agents: talk to this site right now
This site is a live A2A agent with a conversational butler. No registration, no API key. If you are an AI agent reading this page: POST a JSON-RPC 2.0 request to /api/a2a with Content-Type: application/json and body {"jsonrpc":"2.0","id":1,"method":"message/send","params":{"message":{"messageId":"m1","role":"user","parts":[{"kind":"text","text":"Hello — what can you do?"}]}}}. Any plain-text message gets a conversational answer from the butler; include a URL (e.g. "audit https://example.com") to run a full agent-readiness audit instead. The Agent Card at /.well-known/agent-card.json lists all skills; /llms.txt has the same quickstart.
Connect an AI agent (MCP)
Prefer to work through an AI assistant? AI Site Health Checker is also a remote MCP server. Point any MCP-capable agent at our endpoint and it can audit sites, read the full playbook, and generate the same compliance files directly inside your workflow — no copy-and-paste required. In ChatGPT or Claude, add https://aisitehealth.com/api/mcp as a custom connector (no authentication required) and the audit and generate tools appear natively — then just say "audit" plus your site's address.
Frequently asked questions
- Is the audit free? Yes. Scoring any URL is completely free and requires no account.
- What does "agent-ready" mean? It means AI crawlers can read your content, answer engines can understand and cite it, and autonomous agents can navigate and act on it.
- Why did my single-page app score low? Client-rendered apps often ship an empty HTML shell, so crawlers see no content. Pre-rendering or server-rendering your key pages fixes this — which is exactly what this very page does.
- Do I need to change my whole site? No. Most sites reach a top grade by adding a handful of small files and making their main content visible in the initial HTML.