Agent Communication MCP Server

🚀 Cross-System Agent Communication MCP Server

This MCP server enables communication and coordination among different Roo modes/roles, allowing them to collaborate across multiple systems. It establishes an "agent team" architecture, enabling specialized LLM agents to cooperate in task execution, knowledge sharing, and work scheduling.

🚀 Quick Start

Prerequisites

• Node.js 18 or higher
• TypeScript 5.3 or higher
• Authorized access to the GitHub API
• A PlanetScale database account

Installation Steps

1. Clone the repository:

git clone https://github.com/mkc909/agent-communication-mcp-server.git
cd agent-communication-mcp-server

2. Install dependencies:

npm install

3. Create a .env file and add the following variables:

GITHUB_TOKEN=your_github_token
PLANETSCALE_HOST=your_planetscale_host
PLANETSCALE_USERNAME=your_planetscale_username
PLANETSCALE_PASSWORD=your_planetscale_password

4. Build the project:

npm run build

5. Start the server:

npm start

✨ Features

• Agent Registration Center: Register and manage different Roo modes/roles and their capabilities.
• Message Bus: Enable asynchronous communication between agents.
• Task Scheduling: Manage task assignment and progress tracking.
• Knowledge Sharing: Facilitate knowledge transfer among agents.
• GitHub Integration: Create and track GitHub issues, manage pull requests.
• PlanetScale Integration: Store agent data, messages, and tasks in a scalable database.

🔧 Technical Details

The Cross-System Agent Communication MCP server consists of three main components:

1. Core MCP Server

   • Agent Registration Center
   • Message Bus
   • Task Scheduling
   • Knowledge Sharing

2. GitHub Integration Layer

   • Issue Management
   • Pull Request Workflow
   • Project Management

3. PlanetScale Database Layer

   • Agent Data Storage
   • Message Storage
   • Task Database

💻 Usage Examples

Basic Usage

# Run in development mode
npm run dev

# Execute tests
npm test

# Code inspection
npm run lint

📚 Documentation

API Endpoints

Agent Management

• register_agent: Register a new agent.
• update_agent: Update agent information.
• get_agent: Get agent information.
• list_agents: List all registered agents.

Message Communication

• send_message: Send a message to another agent.
• get_messages: Get an agent's messages.
• mark_message_read: Mark a message as read.
• delete_message: Delete a message.

Task Management

• create_task: Create a new task.
• assign_task: Assign a task to an agent.
• update_task_status: Update the task status.
• get_task: Get task details.
• list_tasks: List all tasks.

Knowledge Sharing

• create_context: Create a shared context.
• update_context: Update a shared context.
• share_context: Share a context with an agent.
• get_context: Get a shared context.
• list_contexts: List all shared contexts.

GitHub Integration

• github_create_issue: Create a GitHub issue.
• github_comment_issue: Leave a comment on a GitHub issue.
• github_create_pr: Create a GitHub Pull Request.
• github_review_pr: Review a GitHub Pull Request.

📄 License

MIT