agent_mcp

A robust Python project for managing and interacting with files and directories using Google's ADK and Model Context Protocol (MCP) tools.

Overview

agent_mcp provides an intelligent agent that helps users manage files, list directories, and perform various filesystem operations programmatically. It leverages the google-adk library for agent development and the @modelcontextprotocol/server-filesystem NPM package for advanced filesystem integration.

Features

• List and manage files in a specified directory
• Read and display file contents
• Integrate with Gemini LLM for intelligent file management
• Easily configurable target directory
• Environment variable support for secure configuration

Project Structure

agent_mcp/
│
├── adk_agent_samples/
│   └── mcp_agent/
│       └── agent.py
│
├── mainfoder1/
│   └── rootfolder/
│       ├── __init__.py
│       └── reqerments.txt
│
├── .env
├── .gitignore
├── LICENSE
├── README.md
└── step_By_step.md

Getting Started

Prerequisites

• Python 3.8+
• Node.js and npm
• google-adk
• @modelcontextprotocol/server-filesystem
• python-dotenv for environment variable management

Installation

1. Clone the repository:

   git clone <your-repo-url>
   cd agent_mcp

2. Set up Python environment and install dependencies:

   python -m venv venv
   source venv/bin/activate
   pip install google-adk python-dotenv

3. Install the MCP server filesystem tool:

   npm install -g @modelcontextprotocol/server-filesystem

4. Configure environment variables:

   • Create a .env file in the project root with your Gemini API key:

     GEMINI_API_KEY=your-gemini-api-key
     

Usage

• The main agent logic is in adk_agent_samples/mcp_agent/agent.py.
• By default, the agent manages files in the mainfoder1/rootfolder directory.
• You can modify the TARGET_FOLDER_PATH variable in agent.py to change the target directory.

Example

To run the agent, execute the main script or integrate it into your application as needed:

python adk_agent_samples/mcp_agent/agent.py

Step-by-Step Guide

See step_By_step.md for a detailed setup and usage guide.

License

This project is licensed under the Apache License 2.0.

References