Cookies ahead

Our support chat tool "Intercom" would like to collect some more data on you. See the related link for more details.

Docs

MCP server

🧩

Plug agents into the platform.

The fortrabbit MCP server lets an AI agent read and create apps and environments through the Model Context Protocol, with browser-based authorization instead of a token pasted into a config file.

This article is a draft. The MCP server is in beta and its tool set is still growing.

About the MCP server

The Model Context Protocol is a standard way for AI agents to discover and call the capabilities of an external service. Instead of teaching an agent which HTTP endpoints exist, the agent asks the server what it can do and gets a described list of tools back.

The remote endpoint is https://api.fortrabbit.com/mcp. It is an adapter over the same boundary the public API uses, so authorization and tenant scoping behave identically — an agent never sees more than the person who authorized it.

Connect an agent

Claude Code and Codex both connect over OAuth, so no token has to be handled by hand:

claude mcp add --transport http fortrabbit https://api.fortrabbit.com/mcp
shell
codex mcp add fortrabbit --url https://api.fortrabbit.com/mcp --oauth-client-id https://api.fortrabbit.com/.well-known/oauth-client/codex
shell

The agent opens a browser page that names the requesting client and asks for approval. Approving issues short-lived credentials bound to the agent; the connection can be revoked in the dashboard at any time, which invalidates them immediately.

Available tools

The tool set is deliberately small. Reads cover apps and environments, plus the reference data an agent needs to create one: regions, software presets, PHP versions, component plans, and the connected Git repositories and branches. Writes are limited to creating an app and creating an environment.

Updates, restarts and deletions are not exposed. Adding a tool is a separate decision each time — a REST endpoint existing does not make it available to agents.

MCP or agent skills

The two work together and solve different halves of the problem. The MCP server gives an agent authorized access to platform objects. Agent skills give it the workflow knowledge — how a deployment is done, how a database is synced, which command to run over SSH.

Written by humans and AI in collaboration.

AI use & editorial processEdit on GitHub ↗