Meta-Dynamic MCP Server
Aggregates multiple remote Model Context Protocol endpoints and exposes them through a unified SSE interface, allowing an LLM client to interact with specialized servers without configuration changes.
README
Meta-Dynamic MCP Server
<div align="center"> <strong>🎉 Built with curiousity by the uminai Team</strong> </div>
A single Model Context Protocol (MCP) proxy that aggregates multiple remote MCP endpoints (via HTTP-stream or SSE) and exposes them through one unified SSE interface.
Ideal for driving a single LLM client (e.g. Claude) while mixing in any number of specialized MCP servers (math, finance, etc.).
🔄 Why Meta-Dynamic vs Direct MCP Configuration
Traditionally, you would list each MCP server directly in your LLM client’s mcpServers config. While straightforward, that approach has drawbacks:
- Tight coupling: Every time you add or remove an MCP endpoint, you must update the client config and restart the LLM process.
- Multiple connections: The client has to manage separate HTTP/SSE transports for each server, increasing complexity.
- No shared logic: Common patterns like namespacing, error handling, or retries must be re-implemented in every client.
Meta-Dynamic centralizes these concerns in one proxy:
- Single endpoint: Your LLM client only talks to
http://localhost:8080/sse, regardless of how many backends you add. - Dynamic remotes: Remotes are configured in one place (your proxy), decoupled from the LLM—add/remove without touching the client.
- Unified logic: Namespacing, tool/resource aggregation, error handling, and transport selection live in a single codebase, reducing duplication.
🔧 Prerequisites
- Node.js ≥ v16
- npm (or Yarn)
- A set of running MCP servers you want to proxy (e.g. FastMCP math server on
http://localhost:8083/mcp, CoinGecko’s SSE-based MCP, etc.)
🏗️ Project Structure
meta-dynamic-server/
├── package.json # scripts & dependencies
├── tsconfig.json # TypeScript compiler options
├── .gitignore # Node & dist ignores
├── README.md # this document
└── src/
├── index.ts # bootstrap entrypoint
└── meta-dynamic-server.ts # core proxy implementation
🚀 Installation & Development
-
Clone & install
git clone <repo-url> meta-dynamic-server cd meta-dynamic-server npm install -
Run in watch mode
npm run dev # uses ts-node-dev to reload on changes -
Build & run
npm run build # compiles to `dist/` npm start # runs compiled `dist/index.js`
⚙️ Configuration: Adding Remotes
Edit src/index.ts to define the list of MCP servers you wish to proxy.
Each remote needs:
- name: unique alias (used to namespace URIs & tool names)
- url: full endpoint URL (HTTP-stream endpoints point to
/mcp, SSE to the/ssepath) - transport: either
httpStreamorsse
import { MetaDynamicServer } from "./meta-dynamic-server";
const remotes = [
{ name: "math", url: "http://localhost:8083/mcp", transport: "httpStream" },
{ name: "coingecko", url: "https://mcp.api.coingecko.com/sse", transport: "sse" },
// add more MCP endpoints here
];
new MetaDynamicServer(remotes).start(8080);
Note: The proxy exposes an SSE stream on port 8080 by default:
http://localhost:8080/sse
📜 How It Works
- Remote Initialization: connects to each MCP server using the specified transport.
- Request Handlers:
- resources/list, resources/read → fan-out & namespace by alias
- tools/list, tools/call → aggregate & route tool invocations
- SSE Endpoint: exposes a single SSE stream (
/sse) and message POST path (/messages) for any MCP-capable LLM client.
🧪 Testing
You can verify connectivity with curl or your LLM’s built-in MCP client.
Example with curl to list resources:
# 1. open an SSE stream:
curl -N http://localhost:8080/sse
# 2. in another shell, send a JSON-RPC over POST:
curl -X POST http://localhost:8080/messages \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"resources/list"}'
🚧 Contributing
- Fork the repo
- Create a feature branch
- Submit a PR with tests/documentation
📄 License
Released under the MIT License. See LICENSE for details.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.