MCP Setup

Golden Path uses the Model Context Protocol (MCP) to extend your AI coding assistant with architecture-aware tools. This guide covers how to connect your IDE to the Golden Path cloud MCP server.

What is MCP?

MCP (Model Context Protocol) is an open standard that lets AI assistants call external tools. Golden Path runs as a cloud MCP server that exposes architecture skills as tools your AI can invoke during conversations.

Get your configuration

Sign in to Golden Path Studio and download your mcp.json configuration file. It contains your API key and the server URL, pre-configured and ready to use.

Each API key is scoped to your account and gives the MCP server access to your architectures and skills. See API Keys for how to create and manage keys.

mcp.json

{
  "servers": {
    "golden-path": {
      "type": "http",
      "url": "https://mcp.golden-path.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Claude Code

Create a file called .mcp.json (note the leading dot) inside the .claude/ subfolder of your project root -- not in the project root itself. The full path is your-project/.claude/.mcp.json. Claude Code uses mcpServers as the top-level key.

You can also place it globally at ~/.claude/.mcp.json to make it available across all projects.

After saving, restart Claude Code or start a new conversation. The tools will be available immediately.

.claude/.mcp.json

{
  "mcpServers": {
    "golden-path": {
      "type": "http",
      "url": "https://mcp.golden-path.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

GitHub Copilot (VS Code)

Create a file called mcp.json (no leading dot) inside the .vscode/ subfolder of your workspace root. The full path is your-project/.vscode/mcp.json. VS Code / Copilot uses servers as the top-level key.

Make sure you have the GitHub Copilot extension installed. After saving the config, reload the window.

.vscode/mcp.json

{
  "servers": {
    "golden-path": {
      "type": "http",
      "url": "https://mcp.golden-path.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Cursor

Create a file called .mcp.json inside the .cursor/ subfolder of your project root. The full path is your-project/.cursor/.mcp.json. Cursor uses mcpServers as the top-level key.

After saving, restart Cursor or open a new AI chat session.

.cursor/.mcp.json

{
  "mcpServers": {
    "golden-path": {
      "type": "http",
      "url": "https://mcp.golden-path.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Verify the connection

Once configured, verify the connection by asking your AI assistant:

Prompt

What Golden Path tools are available?

You should see a response listing the platform tools (get_architecture_details, select_architecture, analyze_requirements, etc.) plus any architecture-specific skills.

If the tools don't appear, check that your API key is valid and that your IDE supports HTTP-based MCP servers. You can verify your key in Golden Path Studio.

What's next?

Manage API Keys — Create, revoke, and manage your API keys
Explore platform tools — Learn what each tool does and how to use it
Try the quickstart — Generate your first solution end-to-end