#MCP Smart Inventory Agent This project implements a stateful AI agent using the Model Context Protocol (MCP). It bridges a local FastAPI REST service with an LLM (like Claude Desktop) using LangGraph for request validation and reasoning.

📂 Project Structure

my-mcp-agent/
├── .env                    # API Keys and Secrets
├── docker-compose.yml      # Orchestrates the FastAPI container
├── Dockerfile              # Builds the API image
├── requirements.txt        # Project dependencies
├── api/
│   └── app.py              # FastAPI (REST Endpoints + Auth)
└── mcp_agent/
    ├── mcp_server.py       # MCP Server (The Bridge)
    └── graph_logic.py      # LangGraph (Validation & Logic)

🛠️ Setup Instructions

1. Environment Configuration Create a .env file in the root directory:

INTERNAL_API_KEY=your_secure_token_here
OPENAI_API_KEY=your_openai_key_here  # Required for LangGraph reasoning

2. Launch the REST API Build and run the backend service using Docker:

Bash

docker-compose up -d --build The API will be available at http://localhost:8000. You can view the docs at /docs.

3. Install MCP Dependencies Locally Since the MCP server needs to communicate with your desktop LLM app, install the requirements in a local virtual environment:

Bash

python -m venv venv source venv/bin/activate # Windows: venv\Scripts\activate pip install -r requirements.txt 🔌 Connecting to Claude Desktop To use this agent in the Claude Desktop interface, you must register the server in your configuration file.

Config Location:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

Add this entry:

JSON

{ "mcpServers": { "inventory-agent": { "command": "python", "args": ["/absolute/path/to/my-mcp-agent/mcp_agent/mcp_server.py"], "env": { "INTERNAL_API_KEY": "your_secure_token_here", "OPENAI_API_KEY": "your_openai_key_here" } } } } 🤖 Usage & Features The Interface Once configured, restart Claude Desktop. You will see a 🔌 icon. You can now chat with your API:

Querying: "What is currently in my inventory?"

Smart Adding: "Add 'Industrial Drill' with the description 'High-powered 20V cordless tool'."

LangGraph Validation The agent includes a Validator Node. If you attempt to add an item with a description shorter than 10 characters, the agent will:

Catch the error in the graph_logic.py.

Prevent the API call.

Ask you for more details in the chat window.

🧪 Testing the API Manually If you need to verify the REST endpoints without the LLM:

Bash

This should return 403 Forbidden

curl http://localhost:8000/items

This should return the inventory list

curl -H "X-API-Key: your_secure_token_here" http://localhost:8000/items Would you like me to help you write a Python test script that simulates a full LangGraph run without needing to open the Claude Desktop app?