Your Coding Agent Can Now Ask Pickrate Which Tool to Pick

A week ago we made Pickrate callable by agents inside the browser via WebMCP. Today we're closing the loop: Pickrate is a remote MCP server. Any agent — Cursor, Claude Desktop, a terminal agent, a background workflow — can connect and ask which tool wins, mid-task.

One line to add it

Point any MCP client at the endpoint:

{
  "mcpServers": {
    "pickrate": { "url": "https://pickrate.io/api/mcp" }
  }
}

No API key, no install. Then ask your agent "what's the pick rate for clerk?" or "rank the payments tools agents recommend" and it calls Pickrate and uses the result. Three read tools ship today: getPickRate (one tool's headline number and rank), getLeaderboard (a category ranking), and lookupTool (resolve a name, slug, domain, or package to its report). They return the same live data as the report and leaderboard pages.

Why being callable is the whole point

Pickrate exists to answer one question: when an agent has to choose a tool, does it choose you? For two years the way to win that was to be recommended — named in the answer. But agents increasingly don't just recommend, they act: they call tools, read APIs, and wire up the thing on the spot. At that moment the tools that are callable get used, and the ones that aren't are a sentence the agent has to remember.

This is the end of the Agent Funnel's Discovery stage. Each layer of machine-readability — llms.txt, clean markdown, OpenAPI, MCP — lowers the friction between an agent and a tool. An MCP server removes the last step: the agent doesn't read about Pickrate and then go find an endpoint. It calls Pickrate.

Three surfaces, one source of truth

Pickrate is now reachable three ways, all returning the same numbers: the MCP server (/api/mcp) for agents anywhere, WebMCP (/.well-known/webmcp) for agents in the browser, and a documented REST API (/openapi.json) for everything else. Each one links to the other two, so an agent that lands on any surface can find the rest.

It's also the most honest version of the product. We tell dev tools to be readable and callable by agents. The company measuring that should be the easiest tool in the world for an agent to call.

Add it to your agent, then go see where your tool ranks — and whether an agent would pick it.