Local-MCP-File-Analysis-Server

Manuele-T/Local-MCP-File-Analysis-Server

3.1

If you are the rightful owner of Local-MCP-File-Analysis-Server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

Local-MCP-File-Analysis-Server is a simple MCP server and agent designed to read files, generate summaries, and write reports using STDIO transport.

AI File Analysis Agent with MCP

Cover Image

Overview

This project is a demonstration of the Agent-Tool paradigm using the Machine-to-Machine Communication Protocol (MCP). It features a lightweight AI Agent that orchestrates a series of tasks by communicating with a local Tool Server. The agent's goal is to read a local text file, leverage OpenAI GPT-4.1-Nano API for summarization, and write the resulting summary back to a local summary file.

The application's architecture separates the "brains" of the operation (the Agent) from the "hands" (the Tool). The Agent contains the high-level logic and plan, while the Server exposes specific, self-contained capabilities. Communication between these two components is handled MCP, abstracting away the complexities of network programming.

This project serves as a foundational blueprint for building more complex AI systems where autonomous agents can interact with a variety of digital and real-world tools.

Image

Features

  • AI-Powered Text Summarization: Integrates the OpenAI API to provide context-aware text summaries.
  • Local File System Interaction: The Agent can read from and write to the local file system via the tools provided by the server.
  • Decoupled Agent-Tool Architecture: A clear separation between the agent's logic (agent.py) and the server's capabilities (server.py).
  • Machine-to-Machine Communication: Utilizes MCP for structured communication between the agent and the tool server.
  • External API Abstraction: The complexity of interacting with the OpenAI API is hidden from the agent, exposed as a simple summarize_content tool.
  • Asynchronous Operations: Built on Python's asyncio for high-performance execution.

Tech Stack & Architecture

  • Core Protocol: MCP (Machine-to-Machine Communication Protocol)

    • Server Library: fastmcp (from the mcp package)
    • Client Library: mcp.client modules (from the mcp package)
    • Transport Layer: SSE (Server-Sent Events)

  • Agent (agent.py): The Orchestrator

    • Responsible for the high-level plan and sequencing of tool calls.
    • Connects to the Tool Server using ClientSession and sse_client.
    • Chains inputs and outputs between tools.

  • Tool Server (server.py): The Capability Provider

    • Exposes functions as network-accessible tools using the @mcp.tool decorator.
    • Handles file I/O operations (read_file, write_summary).
    • Manages the integration with the external OpenAI API.

  • AI Core & NLP: OpenAI API

    • Model: gpt-4.1-nano (or any other chat completion model)
    • Task: Natural Language Summarization

  • Language & Libraries:

    • Language: Python
    • Key Libraries: mcp, openai, python-dotenv, asyncio

How to Run This Project

1. Prerequisites

  • Python 3.8+
  • An OpenAI API Key (add it to a .env file)
  • an input.txt file in your project folder

2. Setup

Clone the repository:

git clone <your-repository-url>
cd <your-repository-folder>

Create and activate a virtual environment:

  • On macOS/Linux:
python3 -m venv venv
source venv/bin/activate
  • On Windows:
python -m venv venv
.\venv\Scripts\activate

Install the required packages:

  • With your virtual environment activated, install all dependencies from the requirements.txt file.
pip install -r requirements.txt