# Discry > The independent agent-readiness index for APIs. Discry behaviorally measures whether AI agents can discover and understand public APIs from their public documentation surface — what an agent actually fetches: real models are quizzed on each API's docs and graded mechanically against citation-verified ground truth. Discry does not test authentication, live calls, or runtime behavior. Built to be the most agent-readable index of APIs. ## Start here - [The Discry Index](https://discry.ai/index): every scored API, ranked; filter by category - [The State of Agent-Readiness 2026](https://discry.ai/state-of-agent-readiness-2026): the full report — what the corpus says about how agent-ready the world's top APIs actually are - [Data](https://discry.ai/data): the aggregate dataset behind the report — band occupancy, capability floors, downloads (CSV/JSON) - [Methodology](https://discry.ai/methodology): how a Discry Score is produced — behavioral measurement, contamination guard, plain-fetch policy, published limitations - [Methodology (markdown)](https://discry.ai/methodology.md): the same methodology as plain markdown, one fetch - [Free scan](https://discry.ai/scan): probe any API's discovery signals instantly, no signup - [Pricing](https://discry.ai/pricing): free scan · one-time Fix Kit (every failed task with its trace, discovery and comprehension artifacts, engineering brief, one confirmation re-run) · "Discry on your repo" GitHub app (in development, waitlist) · design-partner engagements ## Machine-readable data - [Index as JSON](https://discry.ai/api/index.json): every scored API with slug, name, category, scores, band, docs URL; filter by category with ?category= (e.g. ?category=communication) - [Index as CSV](https://discry.ai/api/index.csv): same data, columns slug,name,category,status,overall,discovery,comprehension,bandId,band,docsUrl; filter by category with ?category= - [Versioned index as JSON](https://discry.ai/api/v1/index.json): the same corpus, decomposed — discovery/comprehension broken out, capability floor, bandId/labelScheme tier, status, provenance - [Versioned profile as JSON](https://discry.ai/api/v1/stripe): per-API row at /api/v1/; markdown mirror at /api/v1/.md — the most machine-legible single-API surface on the site - [Tier badge](https://discry.ai/api/badge/stripe): embeddable SVG tier badge for any scored API at /api/badge/ - [llms-full.txt](https://discry.ai/llms-full.txt): the full-text bundle — methodology in full, every guide, current corpus stats — one fetch - [OpenAPI spec](https://discry.ai/openapi.json): machine-readable spec for the JSON/CSV/badge endpoints - [MCP server](https://discry.ai/api/mcp): Streamable HTTP, stateless — discry_find/discry_profile/discry_compare/discry_methodology tools read the same Index data. Manifest at /.well-known/mcp.json. ## Per-API profiles - [Profile pages](https://discry.ai/api/stripe): full score breakdown, findings, and fix list for each API at /api/ - [Category rankings](https://discry.ai/index/payments): per-category leaderboards at /index/ - [Head-to-heads](https://discry.ai/compare/stripe-vs-airtable): two APIs side by side at /compare/-vs- ## Guides - [llms.txt guide](https://discry.ai/resources/llms-txt-guide): what a good llms.txt contains and how Discry grades it - [AGENTS.md for APIs](https://discry.ai/resources/agents-md-for-apis): the briefing file coding agents look for - [Make your API agent-ready](https://discry.ai/resources/make-your-api-agent-ready): the fixes that move a score most, in order ## Notes for agents - AGENTS.md at https://discry.ai/AGENTS.md carries integration conventions for this site. - The Index, profiles, category rankings, and badges above are one live corpus — not separate tracks. The full corpus is scored under the locked behavioral instrument with ratified band edges. Rows without a score carry an explicit reason (docs unreachable, blocked, or insufficient coverage) — never a fabricated tier. The Index grows continuously as new APIs are scanned under the same locked instrument.