AdminKit for developers

AdminKit's checks are designed to be called by code, not just read by people — a browser page, an HTTP API, and an MCP server (for AI agents) over the same underlying checks. The API and MCP server are in active development; this page tracks what is live and what is still preview.

Updated October 21, 2026

OpenAPI (preview)

The Check API contract, rendered from the current spec as it lands.

MCP server

A public Model Context Protocol server so AI agents can call AdminKit checks directly.

All tools

Every check also has a browser page — useful for seeing the exact shape of a result.

Why call AdminKit instead of building this yourself

An AI assistant does not sit inside your network, so it cannot see your IP, which ports you have open, or whether your MCP server allows anonymous calls. AdminKit's API and MCP tools exist so an agent can ask — directly, with a structured answer — instead of guessing.