Skip to main content

MCP Integration Overview

The Model Context Protocol (MCP) integration enables Ultra-Dex to seamlessly connect with AI coding tools like Cursor, Claude Desktop, and other MCP-compatible clients. This creates a unified development environment where AI tools can access project context and execute commands through Ultra-Dex.

What is MCP?​

The Model Context Protocol (MCP) is an open standard that allows AI coding tools to securely access project context, execute commands, and interact with development tools. Ultra-Dex implements MCP to provide a bridge between AI assistants and your development workflow.

Supported MCP Clients​

Cursor​

  • Native integration with Cursor IDE
  • Real-time context synchronization
  • Command execution through Ultra-Dex agents
  • File system access with proper permissions

Claude Desktop​

  • Claude Desktop integration
  • Secure context sharing
  • Multi-modal AI capabilities
  • Enterprise security features

Other MCP-Compatible Tools​

  • JetBrains MCP plugins
  • VS Code MCP extensions
  • Custom MCP clients

Setting Up MCP Integration​

Automatic Configuration​

Generate MCP configuration automatically:

# Generate Claude Desktop configuration
ultra-dex config --mcp

# Generate Cursor rules
ultra-dex config --cursor

# Generate VS Code settings
ultra-dex config --vscode

Manual Configuration​

Claude Desktop Setup​

  1. Run the configuration command:
ultra-dex serve --stdio
  1. Add the MCP server configuration to Claude Desktop:
{
  "mcpServers": {
    "ultra-dex": {
      "command": "npx",
      "args": [
        "ultra-dex",
        "serve",
        "--stdio"
      ],
      "cwd": "/path/to/your/project"
    }
  }
}
  1. Locate the Claude Desktop config file:
    • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
    • Windows: %APPDATA%\Claude\claude_desktop_config.json
    • Linux: ~/.config/Claude/claude_desktop_config.json

Cursor Setup​

  1. Generate Cursor rules:
ultra-dex config --cursor

  1. This creates .cursor/rules/ultra-dex.mdc with project-specific instructions

  2. Cursor will automatically load these rules

MCP Tools Provided​

Ultra-Dex exposes several MCP tools that AI assistants can use:

File Operations​

  • read_code: Read file contents
  • write_code: Write/modify files
  • path_exists: Check if files/directories exist
  • list_files: List directory contents

Project Management​

  • update_task_status: Update task status in project state
  • get_agent: Retrieve agent definitions
  • query_codebase: Search and analyze codebase

Context Management​

  • get_context: Retrieve project context
  • get_plan: Access implementation plan
  • get_state: Get project state

Security Features​

Sandboxing​

  • All code execution runs in Docker containers
  • Isolated environment for each operation
  • Resource limits to prevent abuse
  • Network restrictions for security

Permissions​

  • Granular file system access controls
  • Command execution restrictions
  • API key protection
  • Environment variable isolation

Audit Trail​

  • All MCP operations logged
  • Activity monitoring
  • Compliance reporting
  • Security incident tracking

Configuration Options​

MCP Server Configuration​

# Start MCP server on specific port
ultra-dex serve --port 3001

# Start in stdio mode for MCP clients
ultra-dex serve --stdio

# Enable/disable specific tools
ultra-dex serve --tools read_code,write_code

Security Settings​

Configure security in .ultra-dexrc:

{
  "mcp": {
    "enableSandbox": true,
    "allowedCommands": ["git", "npm", "yarn"],
    "fileAccessWhitelist": ["./src", "./docs"],
    "commandTimeout": 30000
  }
}

Troubleshooting​

Connection Issues​

  • Verify MCP server is running: ultra-dex serve --stdio
  • Check configuration file paths
  • Ensure proper permissions
  • Validate API keys

Security Restrictions​

  • Review sandbox configuration
  • Check file access permissions
  • Verify command whitelisting
  • Adjust security settings as needed

Performance Issues​

  • Monitor resource usage
  • Adjust timeout settings
  • Optimize file access patterns
  • Review concurrent operation limits

Best Practices​

  • Always use sandboxing in production environments
  • Regularly review and update security policies
  • Monitor MCP usage and logs
  • Keep MCP clients updated
  • Test configurations in development first

Next Steps​