Fabric MCP Server

What is Fabric MCP Server?

Fabric MCP Server is a standalone server that bridges Daniel Miessler’s Fabric AI framework with the Model Context Protocol (MCP), allowing Fabric’s patterns, models, and configurations to be used directly within MCP‑enabled applications such as IDE extensions or chat interfaces.

How to use Fabric MCP Server?

Install from PyPI with pip install fabric-mcp or uv pip install fabric-mcp. Run fabric-mcp with --stdio (default) for standard I/O transport or --http-streamable for HTTP‑based communication. The server requires a running Fabric REST API instance (fabric --serve). Configure the Fabric base URL, API key, and log level via the environment variables FABRIC_BASE_URL, FABRIC_API_KEY, and FABRIC_MCP_LOG_LEVEL.

Key features of Fabric MCP Server

• Seamless integration of Fabric patterns into MCP clients
• Empowers LLMs in IDEs to use Fabric’s specialized prompts
• Adheres to the open MCP standard for AI tool integration
• Builds on Fabric’s existing CLI and REST API without modifying core
• Exposes MCP tools for listing patterns, running patterns, listing models, and retrieving configuration
• Exposes Fabric pattern templates as MCP prompts for discovery and reuse

Use cases of Fabric MCP Server

• Refactor code inside an IDE using a Fabric pattern invoked via MCP
• Explain code or generate documentation by calling Fabric prompts from a chat interface
• List available Fabric patterns and models directly from an MCP‑aware assistant
• Run a custom Fabric workflow without leaving the editor

FAQ from Fabric MCP Server

What is Fabric?

Fabric is an open‑source framework focused on prompt engineering and modular AI workflows to augment human capabilities.

What is MCP?

MCP (Model Context Protocol) is an open standard that enables AI applications to securely interact with external tools and data sources.

How does Fabric MCP Server work?

An MCP host connects to this server, discovers tools and prompts, then sends requests. The server translates those requests into REST API calls to a running fabric --serve instance, which executes the pattern. The response is streamed back to the host.

What are the prerequisites?

Python 3.10 or newer, and a running Fabric REST API server (fabric --serve). For development, the uv tool is required.

What transport options are available?

The server supports standard I/O (--stdio, default) and HTTP streamable (--http-streamable) transports. For the HTTP transport, you can customize the host, port, and MCP endpoint path.