Discoverability
Discoverability is whether AI agents can discover and understand an API from its public documentation surface. It is a property of what an agent can fetch and parse, distinct from usability, which concerns the experience of working with the API itself. Discoverability determines whether an agent considers an API at all; usability matters only after that choice has already been made.
The distinction carries weight because agents read differently than humans. They fetch over plain HTTP without executing JavaScript and navigate by machine-readable entry points, so documentation that looks excellent to a human in a browser can be effectively invisible to an agent. Discoverability names the layer where that gap lives: the surface as an agent encounters it, before any judgment about how good the content is for a person.
For Discry the distinction is also a boundary. Discry measures discoverability and comprehension because those determine whether agents choose an API; it writes and hosts no documentation and grades every docs stack the same way, platform-hosted or hand-rolled. Usability is real, but it is downstream, and staying out of the docs business is what keeps the grade neutral.
Discry measures discoverability directly: discovery checks probe whether a machine-legible path to the API exists on a plain fetch — llms.txt, AGENTS.md, a parseable spec, MCP surfaces, crawlable sitemaps — and comprehension quizzes then test whether models can operate the API from what that path yields. When discovery fails outright, the profile says so plainly, with the reason and the URLs tried, because a discovery failure is a verifiable property of the public surface and usually the cheapest thing to fix.
What does an AI agent make of your API?
Find out in about a minute — no signup.
Discry your API — free