Box Document Generation using a Pydanthic AI, and the Box MCP Server

A demo showcasing how to build a Pydantic AI agent capable of interacting with the Box MCP server and generating a document.

Overview

This project provides a command-line interface for interacting with Box's document generation features using an AI agent. It enables users to upload templates, process data files, and generate documents automatically through conversational prompts.

Features

• Box Authentication: Same as the MCP server
• File Upload & Management: Upload local files to Box folders
• Document Template Management: Mark files as document generation templates
• Automated Document Generation: Create documents using templates and data files
• AI-Powered Interactions: Converse naturally with the agent to perform tasks

Prerequisites

• Python 3.11 or higher
• Box account with appropriate permissions
• OpenAI API key

Installation

1. Clone the repository:

   git clone https://github.com/box-community/doc-gen-pydantic-ai-box-mcp-server.git
   cd doc-gen-pydantic-ai-box-mcp-server

2. Create a virtual environment:

   uv lock
   uv sync

3. Create a .env file with your OpenAI API keys:

    OPENAI_API_KEY = sk-YOUR API KEY
   

4. In your Box account create a folder named OpenAI Doc Gen

Usage

Run the demo script to see the agent in action:

uv run src/demo.py

This will:

1. Authenticate with Box
2. Show who you're logged in as
3. Upload a template file to a specified Box folder
4. Mark the file as a document generation template
5. Upload data in JSON format
6. Generate a new document based on the template and data

Project Structure

.
├── data/                 # Data files and templates
│   ├── nda_template.docx # Example document template
│   └── NDA.json          # Example data for document generation
├── src/                  # Source code
│   ├── console_utils.py  # Utilities for console output
│   └── demo.py           # Main demo script
├── .env                  # Environment variables (create this)
└── README.md             # This file

How It Works

The agent uses:

• OpenAI's GPT models to interpret user instructions
• pydantic_ai to handle the agent framework
• Box MCP Server to communicate with Box's API

The workflow typically involves:

1. Initializing the agent with access to Box tools
2. Processing user commands in natural language
3. Executing appropriate Box operations
4. Providing feedback on operations performed

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License