home / mcp / blueprint mcp server

Blueprint MCP Server

Diagram generation for understanding codebases and system architecture using Nano Banana Pro.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "arcadeai-blueprint-mcp": {
      "command": "arcade-mcp",
      "args": [
        "deploy"
      ],
      "env": {
        "GOOGLE_API_KEY": "YOUR_API_KEY_PLACEHOLDER"
      }
    }
  }
}

Blueprint MCP enables you to visualize and automate codebase architectures by deploying an MCP server that generates diagrams from your data sources and code. It integrates with Arcade tools to create visual representations of system designs and workflows, helping you understand complex architectures and document them efficiently.

How to use

You run the MCP server from the Arcade command line interface and then access it through your MCP client to start diagram jobs, monitor progress, and retrieve diagrams. Use the server to analyze code, API flows, data pipelines, or any architecture you want to visualize. Start a job, check status, and download the resulting diagram in PNG format or base64 as needed.

How to install

Prerequisites you need before installation are Python, a shell environment, and the Arcade CLI. You will also use a Google AI Studio API key to enable advanced diagram generation.

# Create a Python virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install Arcade MCP package
pip install arcade-mcp

Additional configuration and usage notes

Securely provide your Google AI Studio API key to the Arcade MCP workflow and store it as a secret for use during diagram generation.

arcade-mcp login
arcade-mcp secret set GOOGLE_API_KEY="your_api_key_here"

Available tools

start_diagram_job

Start a diagram generation job. Returns a job ID to track progress.

check_job_status

Check whether a diagram generation job is complete.

download_diagram

Download the generated diagram as a PNG, returned as base64.