Agent Disco Agent discovery, graded.
Agent Disco

Is your site groovy for AI agents?

We check your .well-known/ URIs, your llms.txt, your robots.txt for GPTBot, your OpenAPI spec, and your signup barriers, then grade how easily AI agents can actually use your service.

Why Agent Disco, not just another SEO tool?

  • SEO grades for humans. We grade for agents. Your Core Web Vitals score means nothing to a LangChain tool-use loop or an OpenAI operator. We check the things those runtimes actually read: robots.txt rules for GPTBot & ClaudeBot, /.well-known/ai-plugin.json, llms.txt, OpenAPI specs, machine-readable auth docs.
  • Structured checks with fix hints, not a vibes-based audit. Every grade is a sum of fixed-weight checks you can read at /checks. When something fails, you get a pointer to the standard it's measuring against and a specific remediation. Not "improve your content quality".
  • On-demand & transparent. We don't crawl the web. We only fetch when a human submits your URL. Our scanner is one clearly-identified User-Agent fetching a known short-list of publicly-advertised paths. See /bot for the receipts.

Example scan: stripe.com A · 92

  • crawl.robots_txt Pass · 13/13
  • root_level.llms_txt Pass · 8/8
  • well_known.ai_plugin_json Skip · not set
  • api.openapi_discovery Pass · 10/10

After your first scan

  • Subscribe to grade-change alerts. On any report page, hit Subscribe to grade-change alerts, and we'll email you when a new scan flips the letter grade or moves the score. One mail per scan, no digest, no marketing.
  • Auto-rescan on your cadence. Pick daily / weekly / monthly under Auto-rescan on the report page; we re-scan the host on schedule and track the history without you wiring up CI.
  • Webhooks for CI integrations. Get a signed JSON POST every time a scan of your host completes. Push grades into your status page, fail a build on regression, log to your own dashboard. Setup + payload shape.