LLM Responses MCP Server

by kstrikis
Verified

hybrid server

The server is able to function both locally and remotely, depending on the configuration or use case.

Integrations

  • Used for running the server with commands like 'bun install', 'bun run build', and 'bun run dev'

  • Provides containerization and simplified deployment to EC2 or other server environments with included Docker configuration

  • Used for cloning the repository during deployment to EC2

LLM Responses MCP Server

A Model Context Protocol (MCP) server that allows multiple AI agents to share and read each other's responses to the same prompt.

Overview

This project implements an MCP server with two main tool calls:

  1. submit-response: Allows an LLM to submit its response to a prompt
  2. get-responses: Allows an LLM to retrieve all responses from other LLMs for a specific prompt

This enables a scenario where multiple AI agents can be asked the same question by a user, and then using these tools, the agents can read and reflect on what other LLMs said to the same question.

Installation

# Install dependencies bun install

Development

# Build the TypeScript code bun run build # Start the server in development mode bun run dev

Testing with MCP Inspector

The project includes support for the MCP Inspector, which is a tool for testing and debugging MCP servers.

# Run the server with MCP Inspector bun run inspect

The inspect script uses npx to run the MCP Inspector, which will launch a web interface in your browser for interacting with your MCP server.

This will allow you to:

  • Explore available tools and resources
  • Test tool calls with different parameters
  • View the server's responses
  • Debug your MCP server implementation

Usage

The server exposes two endpoints:

  • /sse - Server-Sent Events endpoint for MCP clients to connect
  • /messages - HTTP endpoint for MCP clients to send messages

MCP Tools

submit-response

Submit an LLM's response to a prompt:

// Example tool call const result = await client.callTool({ name: 'submit-response', arguments: { llmId: 'claude-3-opus', prompt: 'What is the meaning of life?', response: 'The meaning of life is...' } });

get-responses

Retrieve all LLM responses, optionally filtered by prompt:

// Example tool call const result = await client.callTool({ name: 'get-responses', arguments: { prompt: 'What is the meaning of life?' // Optional } });

License

MIT

Deployment to EC2

This project includes Docker configuration for easy deployment to EC2 or any other server environment.

Prerequisites

  • An EC2 instance running Amazon Linux 2 or Ubuntu
  • Security group configured to allow inbound traffic on port 62886
  • SSH access to the instance

Deployment Steps

  1. Clone the repository to your EC2 instance:
    git clone <your-repository-url> cd <repository-directory>
  2. Make the deployment script executable:
    chmod +x deploy.sh
  3. Run the deployment script:
    ./deploy.sh

The script will:

  • Install Docker and Docker Compose if they're not already installed
  • Build the Docker image
  • Start the container in detached mode
  • Display the public URL where your MCP server is accessible

Manual Deployment

If you prefer to deploy manually:

  1. Build the Docker image:
    docker-compose build
  2. Start the container:
    docker-compose up -d
  3. Verify the container is running:
    docker-compose ps

Accessing the Server

Once deployed, your MCP server will be accessible at:

  • http://<ec2-public-ip>:62886/sse - SSE endpoint
  • http://<ec2-public-ip>:62886/messages - Messages endpoint

Make sure port 62886 is open in your EC2 security group!

-
security - not tested
A
license - permissive license
-
quality - not tested

Enables multiple AI agents to share and read each other's responses to the same prompt, allowing them to reflect on what other LLMs said to the same question.

  1. Overview
    1. Installation
      1. Development
        1. Testing with MCP Inspector
          1. Usage
            1. MCP Tools
          2. License
            1. Deployment to EC2
              1. Prerequisites
              2. Deployment Steps
              3. Manual Deployment
              4. Accessing the Server