Catie MCP

Catie MCP

Context Aware Traffic Ingress Engine - A lightweight, configurable proxy service for routing MCP (Model Context Protocol) JSON-RPC requests to appropriate backend services based on request content.

Read the Docs View on GitHub

Dynamic Routing

Route MCP JSON-RPC requests based on method and parameters with pattern-based routing using regular expressions and session-aware routing to maintain client connections.

Streamable Transport

Support for Streamable HTTP transport with SSE (Server-Sent Events) and hot-reloading of configuration without service restart.

Observability

Built-in health check endpoint, Prometheus metrics integration, structured logging, and a monitoring UI with basic authentication.

How It Works

Intelligent Request Routing

MCP Catie analyzes incoming JSON-RPC requests and routes them to the appropriate backend service based on their content. This allows you to distribute your MCP implementation across multiple specialized services while presenting a unified API to clients.

  • Route based on method names and parameters
  • Support for regular expression pattern matching
  • Session-aware routing for consistent client experiences
  • Fallback routes for unmatched requests
resources:
  "^weather/.*": "http://weather-service:8080/mcp"
  "^database/.*": "http://database-service:8080/mcp"
tools:
  "^calculator$": "http://calculator-service:8080/mcp"
  "^translator$": "http://translator-service:8080/mcp"
default: "http://default-service:8080/mcp"
Client
↓
MCP Catie
↙
↓
↘
Weather Service
Calculator Service
Translator Service

Streamlined Architecture

MCP Catie sits between your clients and your backend services, providing a single entry point for all MCP requests. This simplifies client integration and allows you to scale and evolve your backend services independently.

  • Single endpoint for all MCP requests
  • Transparent proxying of requests and responses
  • Support for both synchronous and streaming connections
  • Hot-reloading of configuration without downtime

Comprehensive Monitoring

Keep track of your MCP services with built-in monitoring and observability features. MCP Catie provides detailed metrics and logs to help you understand traffic patterns and troubleshoot issues.

  • Prometheus-compatible metrics endpoint
  • Structured logging with configurable levels
  • Health check endpoint for infrastructure monitoring
  • Web-based monitoring dashboard
mcp_router_requests_total
1,245,678
mcp_router_errors_total
127
mcp_router_response_time_ms
42.3
mcp_router_uptime_seconds
1,209,600

Ready to Get Started?

MCP Catie is designed to be easy to deploy and configure. Follow our documentation to set up your first MCP router in minutes.

Read the Docs View on GitHub

Latest Updates

  1. Understanding MCP Part 1: The Transport Layer

    Understanding MCP Part 1: The Transport Layer. HTTP Streaming and SSE Explained.

  2. MCP Sampling Explained: Best Practices and Examples

    MCP Sampling Explained, Best Practices and Examples

  3. MCP Authentication and Authorization: A Comprehensive Guide

    A comprehensive guide to authentication and authorization in MCP

View All Documentation →