Box MCP
Box MCP Server standardizes context sharing for LLMs, enabling seamless AI access and actions on Box content. It supports remote and self-hosted setups for flexible integration.
How to Install and Use Box MCP
Box MCP is a tool that helps you connect AI agents with content stored in Box. There are two main ways to use Box MCP: the remote server hosted directly by Box, or the self-hosted server you can run yourself. This guide will help you install and start using the self-hosted Box MCP server with simple steps and example code.
Step 1: Clone the Box MCP Server Repository
Before running the server on your local machine, you need to get the code from GitHub. Box MCP is an open-source project, so you can copy it freely.
First, open your command line tool and run this command to download the server:
git clone https://github.com/box-community/box-mcp-server.git
This command copies the Box MCP server code to your computer so you can start setting it up.
Step 2: Install Required Packages
After cloning the code, you need to install the necessary software packages so the server works correctly.
Move into the folder you just downloaded:
cd box-mcp-server
Next, install the dependencies using npm by running:
npm install
This will download all the tools Box MCP needs to run smoothly.
Step 3: Configure Your Environment
Box MCP needs a few setup details before starting. You should create a configuration file to tell the server about your Box account.
Make a file called .env in the root folder and add your Box app credentials. For example:
BOX_CLIENT_ID=your_client_id
BOX_CLIENT_SECRET=your_client_secret
BOX_ENTERPRISE_ID=your_enterprise_id
BOX_PRIVATE_KEY=your_private_key
BOX_PRIVATE_KEY_PASSPHRASE=your_passphrase
Get these details from your Box Developer Console where you created an app. This setup allows Box MCP to safely communicate with your Box account.
Step 4: Run the Box MCP Server
Now that the setup is done, you can start the server. In the command line, run:
npm start
This command will launch the Box MCP server locally on your computer. You should see messages showing that the server is ready.
Step 5: Test the Server with Example Code
To make sure everything works, you can send a simple request to the server. Here is a basic example using curl to list folders:
curl http://localhost:3000/api/folders/0/items
This asks Box MCP for the items inside the root folder (0 is the root folder ID). If set up correctly, you will get back a list of files and folders.
With these steps, you have installed and started the Box MCP server on your local machine. Now AI agents can connect to it and use content from your Box account. For more examples and advanced options, check the official guides on the Box developer site.