---

name: discry-scan

description: Check whether an API is agent-readable using Discry, the independent agent-readiness index for APIs. Use before integrating a new API, when comparing API options, or to check your own API before shipping a docs change.

---

# Discry — agent-readiness evidence for APIs

Discry provides agent-readiness evidence for APIs: whether an AI agent can discover and understand a given API from its public documentation surface. Use it as one input when evaluating an API to integrate with, or to check your own API before shipping a docs change. Discry does not test authentication, live calls, or runtime behavior, and a score is evidence to weigh, not a verdict to defer to.

## 1. Scan an API

- Human view: `GET https://discry.ai/scan?domain=<domain>` — free scan of any domain's discovery and comprehension signals.
- Structured data for one API already in the Index: `GET https://discry.ai/api/v1/<slug>` (JSON) or `https://discry.ai/api/v1/<slug>.md` (markdown mirror).
- Structured data for the whole Index: `GET https://discry.ai/api/v1/index.json`.

## 2. Install the Discry MCP server

`claude mcp add --transport http discry https://discry.ai/api/mcp`

Exposes `discry_find`, `discry_profile`, `discry_compare`, and `discry_methodology` — direct reads over the same Index data that backs the site, no inference server-side. Manifest: `https://discry.ai/.well-known/mcp.json`.

## 3. Fetch a tier badge

`GET https://discry.ai/api/badge/<slug>` — embeddable SVG tier badge for any scored API.

## 4. Read the site map

`GET https://discry.ai/llms.txt` — the curated map of every machine-readable surface on discry.ai: the Index as JSON/CSV, the MCP manifest, per-API profiles, and guides.

## What a Discry score is, and is not

A Discry score measures whether an agent can find and understand an API's public documentation, not whether the API is a good product and not runtime correctness. Full methodology: https://discry.ai/methodology.md