AIScan.site

About

An independent scanner for the agentic web

AIScan grades how well a website works for AI agents and LLM crawlers — the crawlers behind ChatGPT, Claude, Perplexity and Gemini, and the autonomous agents that increasingly browse, read and transact on a user's behalf. Every scan is a set of plain HTTP requests against published, open standards. No opinions, no black box: each result carries the exact URL, status code and body excerpt we based it on.

Who builds it

AIScan is built and maintained by M Asif Rahman, a product engineer working on web infrastructure and SEO tooling. It is not affiliated with, endorsed by, or sponsored by Cloudflare, Vercel, OpenAI, Anthropic or any other vendor whose standards it tests. The scanner is funded by its own Pro plan.

How the rubric is decided

Every check maps to a published specification or a widely-adopted convention — RFC 9309 (robots.txt), the sitemaps protocol, RFC 8288 web linking, RFC 9727 API catalogs, RFC 9457 problem details, llmstxt.org, the Model Context Protocol, OAuth 2.0 protected-resource metadata, and schema.org. We do not invent scoring criteria, and we never grade on aesthetics or marketing copy.

The rubric is versioned. Every scan result records the rubric version used, so historical scores stay explainable when the rubric moves. Checks are tiered into essential (baseline requirements every site should meet), recommended (strong signals most sites benefit from), and bonus (emerging standards that can only add points, never subtract).

What we deliberately don't measure

  • · Search rankings, traffic, or backlinks — that's SEO tooling, not agent readiness.
  • · Anything that requires executing JavaScript in a headless browser. Our scanner is HTTP-only, which keeps scans fast, cheap, and reproducible — but it means client-only rendering will read as thin content, exactly as it does to most crawlers.
  • · Content quality or factual accuracy. We test whether an agent can reach and parse your content, not whether it's good.
  • · Anything behind authentication. We only see what an anonymous agent sees.

Privacy

Scans run against publicly reachable URLs only. Signed-in scans are private by default and you control visibility per scan; anonymous scans are public because they need a shareable report URL. Full detail is in our privacy policy.

Get in touch

Found a check that's wrong, or a standard we should support? Contact us — we read everything. Developers should start at the developer guide.