Pick Rate API

Pull any tool's Pick Rate, rank a category, run your own evals, or let an agent call Pickrate directly. Three ways in: a public read API, an authenticated eval API, and an MCP server.

Public read API

No key required. Both endpoints return JSON.

Get a tool's Pick Rate

GET https://pickrate.io/api/public/report?q=<slug|name|domain|package>

q is anything that identifies a tool — stripe, clerk, @sendgrid/mail. Returns the tool's Pick Rate, rank, and category. 404 if we don't track it yet.

Rank a category

GET https://pickrate.io/api/public/leaderboard?category=<slug>

Returns the ranked vendors for a category (e.g. payments, auth, search). Omit category to list the available categories.

Full schema: openapi.json (OpenAPI 3).

Authenticated eval API

Run and read your own evals programmatically. Mint a key in Settings (paid plans) and send it as a bearer token.

Authorization: Bearer pk_live_…

GET https://pickrate.io/api/v1/evals             # your evals
GET https://pickrate.io/api/v1/runs/:runId       # a run's scores
GET https://pickrate.io/api/v1/runs/:runId/traces # full per-trial trace data

curl -H "Authorization: Bearer pk_live_…" \
  https://pickrate.io/api/v1/evals

MCP server

A remote Model Context Protocol server (Streamable HTTP) so an agent can query Pickrate mid-task. No key required.

https://pickrate.io/api/mcp

Tools:

• getPickRate — a tool's Pick Rate by name, slug, domain, or package.
• getLeaderboard — ranked vendors for a category.
• lookupTool — resolve a search term to a tracked tool.

Listed on the MCP Registry as io.pickrate/pickrate. The links these tools return are tagged so an agent-driven click is attributable — the discovery side of the same funnel the attribution product measures.

Other machine surfaces

• llms.txt — the site map for LLMs.
• Append .md to most pages (leaderboards, reports, methodology) for clean Markdown.

Questions

← All docs · Agent Attribution →