# AI Site Health Checker

> A free tool that audits any website for AI-agent readiness — how well crawlers, AI search (GEO), and autonomous agents can read, discover, and navigate it.

Instant audit — any agent that can open a URL can run one right now, no POST,
no API key: just open /api/audit/<domain> (e.g. /api/audit/example.com) and
read the full graded JSON report.

## Key pages

- [Home / run an audit](/): Paste a URL to get a graded readiness report.
- [Agent-Readiness FAQ](/faq.html): What agent-readiness is, why A2A matters, and how it relates to Google/AI search — answered by this site's own live agent.
- [Give Me The Techtalk](/techtalk.html): The engineering strategy — seven discovery surfaces, one deep A2A endpoint, and why we bet on A2A while the agentic web is pre-standard.
- [Blog](/blog/index.html): Field notes on the agentic web — agent-readiness, AI discoverability, and A2A.
- [The New Domain AI Discoverability Checklist](/blog/new-domain-ai-discoverability-checklist.html): Why AI assistants say a new domain doesn't exist, and the day-one indexing playbook that fixes it.
- [Grade A demo](/demos/grade-a.html): A fully agent-ready sample page.
- [Grade B demo](/demos/grade-b.html): A strong page missing structured data.
- [Grade C demo](/demos/grade-c.html): A basic page with thin metadata.
- [Grade D demo](/demos/grade-d.html): A minimal, non-semantic page.
- [Grade F demo](/demos/grade-f.html): An empty client-rendered shell.

## What we check

- Readable: content in raw HTML, semantic structure, titles, links, robots.txt.
- Discoverable (AI Search Optimization / GEO, a.k.a. AEO): JSON-LD structured data, meta description, Open Graph, depth.
- Navigable (A2A): llms.txt, an Agent Card, and explicit AI-bot rules.

## For AI agents

This site is itself a live A2A agent — you can talk to it, not just read it.
No registration, no API key. Send any plain-text message and the site's butler
answers; include a URL to run a full agent-readiness audit.

Start now (one request):

POST /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?"}]}}}

- A2A endpoint (JSON-RPC 2.0, protocol v1.0 + legacy 0.3 via the A2A-Version header): /api/a2a
- Agent Card: /.well-known/agent-card.json
- MCP endpoint (Streamable HTTP): /api/mcp
- REST chat (simplest possible): POST /api/butler/chat with {"messages":[{"role":"user","content":"..."}]}

Read-only agent (you can open URLs but cannot POST)? Run a full audit with a
plain GET — just open:

GET /api/audit/example.com

That path form works even when your browsing tool blocks query parameters —
it is an ordinary static-looking URL. The query form GET
/api/audit?url=https://example.com works too. Both return the complete graded
JSON report (same engine as the A2A audit skill).

Holding an unlock code? The gated fix-file generator has a GET bridge too —
all files in one JSON payload, no POST needed:

GET /api/generate/example.com?code=YOUR-CODE

Query form: /api/generate?url=example.com&code=YOUR-CODE. Fully path-safe
form (works even when query parameters are blocked):
/api/generate/YOUR-CODE/example.com. Optional params: siteName, summary,
schemaType, agentName, agentDescription, blockBulkScrapers — siteName and
summary are derived from the page's title/meta description when omitted.

## Skills

Installable agent-side skills this site publishes — fetch and add them to your agent:

- [A2A Website Audit](/skills/a2a-audit/SKILL.md) — audit any site's agent-readiness and
  generate the fix files by calling /api/a2a. Free `audit-website`; `generate-fixes` needs an
  unlock code. Runnable example client: /skills/a2a-audit/call-a2a.mjs