AI Agent Readiness Scanner

Is your website ready for AI agents and AI search? This Actor scores any website from 0 to 100 on how well it can be found, read and used by ChatGPT, Claude, Perplexity, Gemini and autonomous agents, and tells you exactly what to fix first.

Classic SEO checks don't cover the agentic web. This Actor checks the new standards:

• llms.txt / llms-full.txt: present, well formed (H1, summary, sections, links)
• AI crawler access: robots.txt rules for 18 AI bots, grouped by purpose: training (GPTBot, ClaudeBot, Google-Extended, CCBot...), AI search (OAI-SearchBot, Claude-SearchBot, PerplexityBot...) and user-triggered fetchers (ChatGPT-User, Claude-User, Perplexity-User...)
• Content-Signal (search, ai-input, ai-train) and noai meta / X-Robots-Tag
• Content readability: how many words are visible without JavaScript (most AI crawlers don't run JS), schema.org JSON-LD types, H1, language, Markdown alternates
• Agent interfaces: live MCP servers (/mcp, mcp.<domain>), MCP Server Cards, AI Catalog (/.well-known/ai-catalog.json), A2A Agent Cards (/.well-known/agent-card.json), OpenAPI specs, OAuth authorization server metadata, WebMCP (navigator.modelContext) and NLWeb

Scoring

Four pillars of 25 points each: Discoverability, AI Access, Content Readability and Agent Interfaces. You get an A–F grade and a level: Invisible to AI, Partially visible, AI-search ready or Agent-native.

Every site also gets prioritized recommendations (high / medium / low impact) and a single top fix.

Use cases

• SEO / GEO agencies: run bulk AI-visibility audits for prospects and clients. A bad grade is an easy pitch.
• Lead generation: find companies that block AI search bots or have no llms.txt, then sell them the fix.
• Brands and publishers: check you aren't accidentally blocking AI search while trying to block training.
• Market research: track how fast a sector adopts MCP, A2A and llms.txt.
• Monitoring: schedule it weekly and get alerts when someone changes robots.txt.

Input

{ "websites": ["stripe.com", "nytimes.com", "yourclient.com"] }

Output example

{
  "domain": "stripe.com",
  "grade": "B",
  "score": 76,
  "level": "Agent-native",
  "pillars": {"discoverability": 22, "aiAccess": 23, "contentReadability": 19, "agentInterfaces": 12},
  "summary": "stripe.com: B (76/100) Agent-native. llms.txt yes, AI search bots 7/7 allowed, MCP yes, A2A no, JSON-LD 7 types",
  "llmsTxt": {"present": true, "hasH1": true, "links": 120, "llmsFullTxt": false},
  "aiCrawlerPolicy": {"blockedBots": [], "byPurpose": {"train": {"allowed": 7, "total": 7}, "search": {"allowed": 7, "total": 7}}},
  "agentInterfaces": {"mcpServers": [{"url": "https://mcp.stripe.com/", "authRequired": true}], "a2aAgentCard": null, "openApi": null},
  "recommendations": [{"impact": "low", "recommendation": "Add /llms-full.txt ..."}]
}

There are two dataset views: Overview (one row per site) and Recommendations (one row per fix). Export to CSV, Excel or JSON, or send the results to Sheets, Slack, HubSpot, n8n, Make or Zapier.

Pricing

Pay per event: you are charged once per website successfully scanned. Unreachable sites are free.

How it works

The scanner makes about 15 lightweight HTTP requests per site: the homepage, robots.txt, llms.txt, well-known files and an MCP initialize probe. It doesn't need a browser or proxies, and it never calls MCP tools or sends form data.

FAQ

Should I block AI bots? That's up to you, but block training bots and search bots separately. Blocking OAI-SearchBot or PerplexityBot means you won't be cited in AI answers.

What is llms.txt? A Markdown file at /llms.txt that gives LLMs a curated map of your site.

What is an A2A Agent Card? A JSON file at /.well-known/agent-card.json that describes an AI agent you run, as defined by the Agent2Agent protocol.