DeepSeek · Ai◎ discry
Legible
Usable, with real gaps
DISCOVERY31.6
COMPREHENSION89.9
Small-model ready#31 of 51 Ai APIs · 41st percentile
AI · RANK #31 OF 51

DeepSeek is Legible to agents.

Discry independently scored how well an AI agent can discover and understand the DeepSeek API from what’s public — measuring discovery and comprehension separately, not a single blended number.

Familiarity 86% · what a model knew before reading the docs

Discovery

32 / 100
OpenAPI specfound at https://api-docs.deepseek.com/openapi.json but invalid: unparseablePartial
llms.txtmissing headings, links, or descriptionsPartial
llms-full.txt878 tokensPass
AGENTS.mdnot found in repoFail
.well-known/mcp.jsonnot presentFail
MCP registrysmithery match for "deepseek"Pass
robots.txt AI directivesno robots.txtFail
Sitemapnot presentFail

Comprehension

90 / 100
Factual recall2/2 passed
Request construction1/1 passed
Absence honesty1/1 passed

See your own Discry result.

Drop your API docs URL. See what an agent sees — in 60 seconds, free.

Discry your API — free