Lesson 1: MCP Server Setup

What Is MCP?

The Model Context Protocol (MCP) is a standardized way for AI tools to connect to external data sources and services. With MCP, AI editors like Cursor, Copilot, and others can query Sruja's architecture graph directly.

Why MCP for Sruja?

Without MCP, AI editors don't know your architecture:

• AI suggests changes without understanding dependencies
• AI doesn't know which components are critical
• AI can't tell you the blast radius of a change

With MCP, AI editors can:

• Query architecture graph in real-time
• Understand component relationships
• Provide architecture-aware suggestions
• Warn about breaking changes

Setting Up Sruja MCP

1. Install Sruja CLI

# From source
cargo install --path crates/sruja-cli

# Or from release
curl -fsSL https://sruja.ai/install.sh | bash

2. Verify MCP is Available

sruja mcp --help

3. Configure for Cursor

Create or edit ~/.cursor/mcp.json (or .cursor/mcp.json in your project):

{
  "mcpServers": {
    "sruja": {
      "command": "sruja",
      "args": ["mcp", "-r", "."],
      "env": {}
    }
  }
}

4. Configure for VS Code (Copilot)

// .vscode/mcp.json in your project
{
  "sruja": {
    "command": "sruja",
    "args": ["mcp", "-r", "."]
  }
}

5. Configure for Cline/Windsurf

// .clinerules or .windsurf/rules/sruja.md
{
  "mcpServers": {
    "sruja": {
      "command": "sruja",
      "args": ["mcp", "-r", "."]
    }
  }
}

Testing the Connection

# Start MCP server in test mode
sruja mcp -r . --test

# Or start and watch logs
sruja mcp -r . --verbose

In your AI editor, try asking:

• "What components does the API depend on?"
• "Show me the architecture context"
• "What would break if I change the database?"

MCP Tools Available

Once connected, the AI editor can use these Sruja tools:

Next Steps

Lesson 2 covers building and optimizing context for AI assistance.