Set Up Vega DevTools MCP Server

Vega DevTools MCP is a local Model Context Protocol (MCP) server that installs on your development machine and provides Vega-specific tools and context directly into AI agents like Cursor, Amazon Q, or GitHub Copilot. The server gives your AI agent direct access to Vega-specific knowledge, so you can get accurate answers about Vega development, run diagnostic commands, and follow Vega best practices without leaving your coding environment.

Prerequisites

1. Install Node version 20 or 22 (24 isn't supported).

   While MCP works with Node.js 18, avoid using it since Node.js 18 reached it's end-of-life in April 2025.

2. Vega SDK installed on your machine.

3. An AI agent.

   Vega DevTools MCP works with all AI agents that support MCP. Click each AI agent name to view the setup instructions:

   • Cursor
   • GitHub Copilot
   • Claude Code CLI
   • Kiro
   • Cline
4. (Optional) A Vega app project.

Install and configure the MCP server

1. Add the following configuration to your AI agent's MCP settings file:

   Copy code Copied to clipboard.

   "vega-devtools-mcp": {
     "type": "stdio",
     "command": "npx",
     "args": [
       "-y",
       "@amazon-devices/vega-devtools-mcp@latest"
     ]
   }
   

2. Run the following command to verify the installation of the MCP server:

   Copy code Copied to clipboard.

   npx @amazon-devices/vega-devtools-mcp@latest --help
   

   Expected output:

   🚀 Vega Developer MCP Server
   ============================
   Vega Developer MCP Server provides Vega development tools and capabilities.
   CONFIGURATION:
   To use this MCP server with your AI client, add the following configuration:
   For Amazon Q (/Users/username/.aws/amazonq/mcp.json):
   {
     "mcpServers": {
      "vega-devtools-mcp": {
        "command": "npx",
        "args": [
          "-y",
          "@amazon-devices/vega-devtools-mcp@latest"
        ],
        "type": "stdio"
       }
     }
   }
   USAGE:
    npx @amazon-devices/vega-devtools-mcp@latest --init-context     Initialize Vega context for AI agents
    npx @amazon-devices/vega-devtools-mcp@latest --help             Show this help message
    npx @amazon-devices/vega-devtools-mcp@latest —-version          Show version information
   

3. Verify Vega DevTools MCP is installed in your AI Agent.

   In your AI Agent's chat interface, run this prompt:

   Copy code Copied to clipboard.

   List the tools provided by Vega DevTools MCP
   

   You should see a response that includes the following tools:

   • analyze_perfetto_traces
   • read_document
   • list_documents

4. Initialize project context.

   This step directs the AI Agent to use vega-devtools-mcp for Vega-related queries. Skipping this step results in read_document error.

   4.1 Navigate to your Vega app project directory and run:

   Copy code Copied to clipboard.

   npx @amazon-devices/vega-devtools-mcp@latest --init-context
   

   4.2 Select your AI agent from the list:

    ====================================================
    🚀 Vega Developer Tools - Initialize Vega Context
    ====================================================
    This tool installs Vega App development context document in your project.
    The context document guides AI agents to efficiently answer queries related to app development for Vega operating system (OS).
    Choose from the following options to automatically install agent-specific preset in your app project:
    1. Cursor - Adds AGENTS.md file
    2. Claude Code - Adds CLAUDE.md file
    3. GitHub Copilot - Adds AGENTS.md file
    4. Amazon Q - Adds .amazonq/rules/ directory
    5. Kiro - Adds .kiro/steering/ directory
    6. Cline - Adds .clinerules/ directory
    7. Other AI agent or custom setup - View content only
    Select an AI agent (1-7):
   

   After you select your AI agent, Vega DevTools MCP automatically creates the appropriate context file in your project directory:

   AI Agent	Context File Location
   Cursor	project-root/AGENTS.md
   Claude Code	project-root/CLAUDE.md
   GitHub Copilot	project-root/AGENTS.md
   Amazon Q	project-root/.amazonq/rules/
   Kiro	project-root/.kiro/steering/
   Cline	project-root/.clinerules/

   Note: If these context files don’t properly load for you, enter “use AGENTS.md for context” at the start of any AI agent session

5. Run this prompt in your AI agent to test the configuration:

   Copy code Copied to clipboard.

   Can you describe the React Native for Vega Platform Architecture in one sentence?
   

   The AI agent invokes the read_document MCP tool from Vega DevTools MCP. Grant permission if prompted.

   Expected response:

   Vega is a TV platform that uses a system-bundled React Native runtime (rather than app-bundled), where applications only package their JavaScript code and dynamically link to the operating system (OS)-provided React Native framework and native services at runtime for optimized performance and resource sharing.

Use the available tools

After you configure the MCP server and initialize context, you can use the following Vega-specific tools and prompts in your AI agent to get help with Vega development tasks.

MCP tools

Use these tools to access Vega documentation and analyze your app's performance:

MCP prompts

If your AI agent support MCP prompts, MCP prompts are directly invocable in your AI agent. For example, in VS Code GitHub Copilot, you can invoke MCP prompts using a slash command.

Example prompts

Add a home screen component to vega app:

For this Vega app, I want to create a new component called HomeScreen with just some placeholder text of "Home Screen". And I want to replace the content of App.tsx with HomeScreen and it should take up the entire screen.

Add a stack navigator in vega app:

Create a new Vega stack navigator for my project. The Home Screen should be the default (and only) route for now.

Question related to cli commands:

What cli command should I use to monitor application activity and performance metrics?

List all capabilities:

Can you list all documents related to feature development on vega?

Troubleshooting

If you encounter any issues, see Troubleshoot Vega DevTools MCP.

Related topics

• Build Your App
• Run Your App on Vega Virtual Device or Fire TV Stick

Last updated: Dec 22, 2025