Consulting Agents MCP

🚀 Consulting Agent Service Documentation

This documentation provides a comprehensive guide to the installation, configuration, usage, and troubleshooting of the consulting agent service. It also includes information on project structure, licensing, and contact details.

🚀 Quick Start

Before you start using the consulting agent service, make sure you have met the installation requirements and completed the configuration steps. Then, you can start the service and use the HTTP API to call different models.

✨ Features

• Multiple Models: Supports different models for various types of consulting needs, such as basic consultation, technical search, and code analysis.
• HTTP API: Allows easy integration with other systems through HTTP requests.
• Configuration Flexibility: Configurable through environment variables for different transport protocols, host, and port settings.

📦 Installation

Environment Requirements

• Operating System: Unix-based systems (e.g., Linux or macOS).
• Python Version: 3.6 or higher.
• Dependency Management Tool: pip.

Installation Steps

1. Clone the project repository:

   git clone https://github.com/username/repo.git
   cd repo
   

2. Install the dependencies:

   pip install -r requirements.txt
   

📚 Documentation

Configuration Guide

Environment Variable Configuration

Create a .env file in the project's root directory and add the following content:

MCP_TRANSPORT=stdio  # Optional values: stdio, http, sse. Default is stdio.
HOST=127.0.0.1        # Default value: 127.0.0.1
PORT=5000            # Default value: 5000

Usage Guide

Start the Service

DEBUG=true ./start_mcp_server.sh

Call the Model via HTTP API

When using the HTTP transport protocol, you can call different models through the following interfaces:

• Darren/Sonny: For basic consultation questions.
• Sergey: For searching related technical questions.
• Gemma: For code analysis based on GitHub repositories.

Example Request

{
  "agent": "Darren",
  "consultation_context": "I encountered a code error, where...",
  "source_code": "def example():\n    return 'hello'"
}

Troubleshooting

Frequently Asked Questions

1. MCP server not found: Check if the absolute path in the claude mcp add command is correct.
2. API authentication failed: Confirm that the API key in the .env file is set correctly.
3. Connection issues: Make sure the MCP server is started before running Claude Code.

Debug Logs

View the detailed logs in the MCP server terminal for more debugging information.

Update Guide

1. Pull the latest code:

   git pull origin master
   

2. Restart the MCP server:

   ./start_mcp_server.sh
   

Development Guide

Run the Development Version

DEBUG=true ./start_mcp_server.sh

Test the Interface

Use the curl command to test the HTTP interface:

# Test the Darren model
curl -X POST http://localhost:5000/consult \
  -H "Content-Type: application/json" \
  -d '{"agent":"Darren","consultation_context":"Test message"}'

Project Structure

• mcp_consul_server.py: Core implementation of the MCP server.
• start_mcp_server.sh: Startup script.
• requirements.txt: Dependency management file.

📄 License

Please refer to the LICENSE file in the project.

📞 Contact

If you have any questions, please contact support@consulting-agents.com.