Developer Portal

MCP Introduction

The Model Context Protocol (MCP) endpoint exposes Admetrics Query Language (AQL) analytics as validated tools that agents and coding assistants can call directly. Use it when a chat product or coding agent should search the catalog, compile natural language, and run structured analytics queries without writing HTTP code.

When To Use MCP

  • Connect ChatGPT, Claude, Codex, OpenCode, and other MCP clients to catalog-backed analytics.
  • Let an agent discover exact metric, dimension, and client IDs before running a query.
  • Keep validation-before-execution guarantees inside the tool layer.

Use REST instead when a backend service or job needs explicit HTTP contracts. See the REST API Reference.

Set Up A Client

  • Chat App Setup — connect chat products such as ChatGPT and Claude with OAuth.
  • MCP Setup — configure coding agents and servers with a config file or CLI.
  • MCP Reference — the advertised tools, transport, and contract details.

Endpoint

POST https://api.admetrics.io/mcp

Streamable HTTP transport. Authenticate with OAuth where the client supports it, or a managed bearer API key. Managed keys map to owner permissions but have independent lifecycle, limits, and usage reporting.