MCP Fabric Documentation

MCP Fabric is an enterprise-ready, high-performance adaptive runtime fabric and intelligent load balancer for the Model Context Protocol (MCP). Inspired by high-availability distributed systems like PyTorch Distributed and Kubernetes mesh, MCP Fabric brings dynamic topology management, session affinity, health monitoring, zero-downtime routing, and multi-agent scale-up capabilities to MCP server fleets.

Getting Started

• Installation
  • Python Package
  • TypeScript / Node.js Workspace
  • Requirements
• Quickstart Guide
  • 1. Launching the Adaptive Fabric Gateway
  • 2. Connecting via Python SDK
  • 3. Inspecting Pool Telemetry
• Architecture Overview
  • Architecture Design

Core Concepts & Architecture

• Adaptive Runtime Fabric
  • Adaptive Runtime Fabric Decisions
• Load Balancing Architecture
  • Load-Aware Routing
• Session Affinity & State Management
  • Session Affinity
• Mode Recovery Matrix
  • Mode Recovery Matrix

Tutorials & Guides

• Python SDK Tutorial
  • Basic Usage
• TypeScript SDK Tutorial
• Standalone Gateway Guide
  • Standalone Gateway Packaging
• Deployment Guide
  • Deployment Guide

API Reference

• Python API Reference
  • Local Fabric Context
  • Fabric Client
  • Runtime Gateway Manager
• TypeScript API Reference
  • Core Session Registry
  • Load Router
• CLI Reference
  • Commands

Indices and Tables

• Index

• Module Index

• Search Page