Free scan · no e-mail asked

Can AI assistants read your website?

Enter your domain. The scan reads the public pages an AI agent would read and gives you a score out of 100, with what we saw and what to change for each check.

What the scan checks

9 checks. 4 are essential and share 80 points; 5 are recommended and share 20. A partial result earns part of its points.

  • Home page available as Markdown

    Essential · 20

    We ask your home page for Markdown ("Accept: text/markdown"). An agent reads a clean Markdown page faster and more reliably than HTML full of layout.

  • Real 404 for agents

    Essential · 20

    We request a made-up URL. A real 404 tells an agent the page does not exist; a 200 "soft 404" makes it believe the page is real. A short Markdown 404 that points to llms.txt helps it find its way back.

  • llms.txt present

    Essential · 20

    We read /llms.txt: a plain-text file, with a "# " title and a "> " summary, that tells AI assistants what your business is and which pages matter.

  • Content readable without JavaScript

    Essential · 20

    We read the raw HTML of your home page, without running JavaScript, as most AI crawlers do: a title, enough text, and headings in order.

  • llms.txt says when to use you and how to contact you

    Recommended · 4

    We look in llms.txt for a "when to use" section (the requests you are the right answer for) and a "contact / book" section (how an agent should reach you).

  • robots.txt states a policy for AI crawlers

    Recommended · 4

    We read /robots.txt: it must not block the whole site, and it should state what you decide for AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended). Blocking them is your choice: we report it, we do not mark it down.

  • XML sitemap

    Recommended · 4

    We look for an XML sitemap, declared in robots.txt or at /sitemap.xml, so agents can find every page.

  • Organization data (JSON-LD)

    Recommended · 4

    We look on your home page for JSON-LD describing your organization, with a contact point and a postal address.

  • About, contact and privacy pages

    Recommended · 4

    We follow the links of your home page to an about page, a contact page and a privacy policy, each with real text.

How the scan works

  • Only GET requests to public pages, at most 12 per scan, each limited to 5 seconds and 1 MB.
  • HTTPS only. No login, no form, no port scan, no security test.
  • Your domain and the result are not kept. Your IP address is only held briefly in memory, to limit the number of scans.
  • Your results stay private: nothing is published.
  • The requests come from DevForge-AgentScan/1.0. If you see it in your logs, someone asked for a scan of your site on this page.