Agent resource

A read-only MCP endpoint for agentic commerce evidence

Agents can query the site’s protocol registry, documented adoption records, public evidence and readiness criteria through a small, read-only MCP server.

Scope: the server does not store data, accept credentials, scan merchant URLs, initiate transactions or provide protocol-owner certification. It only returns information already published by this site and preserves source URLs in the response.

Endpoint

https://agenticcommerceprotocol.info/mcp

Example configuration

{
  "mcpServers": {
    "agenticCommerceProtocolInfo": {
      "url": "https://agenticcommerceprotocol.info/mcp"
    }
  }
}

This is a generic remote-MCP configuration shape. Confirm the exact configuration syntax in your client’s documentation before adding a remote server. The endpoint does not require credentials, so a client that asks for a key is not using this server’s documented setup.

Versioned protocol handshake

Server version 1.1.0 implements the 2025-06-18 MCP lifecycle for a read-only tools surface. Send initialize first, then include MCP-Protocol-Version: 2025-06-18 on every later HTTP request.

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "initialize",
  "params": {
    "protocolVersion": "2025-06-18",
    "capabilities": {},
    "clientInfo": {
      "name": "example-client",
      "version": "1.0.0"
    }
  }
}

Available tools

Tool	Purpose
`list_protocols`	Return each covered protocol, scope, maturity and source-review date.
`get_protocol`	Return one protocol with documented features, adoption rows and primary sources.
`get_adoption_records`	Return source-linked organization records, optionally filtered by protocol.
`get_evidence`	Return public evidence records, optionally filtered by a claim ID or component.
`get_readiness_criteria`	Return the transparent self-assessment criteria and their weights.

Operating limits and observation

The endpoint accepts no browser cross-origin calls, requires no session and exposes no write capability. Cloudflare request logs and the scheduled public-endpoint check are used to investigate transport errors. Raw data is also available through the evidence JSON and Observatory JSON.