Implement Model Context Protocol (MCP) Runtime Support for AI Agent Integration

[Copilot]

Why make this change?

• Closes [Enh]: Add MCP endpoint /mcp #2832
• Enables AI agents and applications to interact with Data API Builder through the standardized Model Context Protocol (MCP), providing a dedicated /mcp endpoint for AI-driven automation and orchestration.
• Allows agents to discover available entities, perform CRUD operations with entity-aware validation, and execute stored procedures in a way that models and agents can reliably consume.

What is this change?

This PR implements comprehensive MCP (Model Context Protocol) support for Data API Builder, adding:

Core MCP Runtime Configuration

• New McpRuntimeOptions class with configurable endpoint path, enabled/disabled state, and DML tools configuration
• JSON schema updates to support MCP configuration in dab.draft.schema.json
• CLI integration with new options for dab init and dab configure commands:
  • --mcp.enabled / --runtime.mcp.enabled
  • --mcp.path / --runtime.mcp.path
  • DML tools configuration options (describe-entities, create-record, read-record, update-record, delete-record, execute-record)

MCP Server Implementation

• Enhanced MCP server setup with proper service provider integration and conditional configuration
• DML tools integration that respects configuration settings for individual operations
• Async/await improvements for better performance and reliability
• Development debugging support with automatic debugger attachment

Configuration Integration

• Runtime options integration - MCP joins REST and GraphQL as a first-class API endpoint type
• Path validation using existing URI validation infrastructure
• Health endpoint support - MCP status included in health checks
• Base route compatibility - MCP works with configured base routes

Testing & Validation

• Comprehensive test coverage across 25+ test files updated
• Configuration validation tests ensuring MCP settings are properly validated
• Health endpoint tests including MCP status validation
• Integration tests for MCP endpoint enabling/disabling scenarios

How was this tested?

• Integration Tests - Updated existing integration tests to include MCP configuration
• Unit Tests - Added comprehensive unit tests for MCP configuration, validation, and runtime behavior

Sample Request(s)

CLI Configuration

# Initialize with MCP enabled
dab init --database-type mssql --connection-string "..." --mcp.enabled true --mcp.path "/mcp"

# Configure MCP settings
dab configure --runtime.mcp.enabled true
dab configure --runtime.mcp.path "/api/mcp"
dab configure --runtime.mcp.dml-tools.enabled true

Configuration JSON

{
  "runtime": {
    "mcp": {
      "enabled": true,
      "path": "/mcp",
      "dml-tools": {
        "enabled": true,
        "describe-entities": true,
        "create-record": true,
        "read-record": true,
        "update-record": true,
        "delete-record": true,
        "execute-record": true
      }
    }
  }
}

MCP Request Example

# List available MCP tools
curl -X POST http://localhost:5000/mcp \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/list"
  }'

This implementation provides the foundation for AI agent integration with Data API Builder, enabling automated database operations through the standardized Model Context Protocol.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.