AnilPuram opened a new pull request, #18143:
URL: https://github.com/apache/druid/pull/18143
### Description
This PR introduces a new **Apache Druid MCP Server** that enables AI agents
and applications to interact with Apache Druid through the Model Context
Protocol (MCP). The MCP server provides a standardized interface for AI tools
like Claude, ChatGPT, and other MCP-compatible clients to query Druid
datasources, retrieve metadata, and perform administrative operations.
#### What is MCP?
The Model Context Protocol (MCP) is an open standard that enables AI
applications to securely connect to data sources and tools. By providing an MCP
server for Apache Druid, we're making Druid more accessible to the growing
ecosystem of AI-powered applications and workflows.
#### Added MCP Server Implementation
- **Location**: `mcp-server/` directory containing a complete
TypeScript-based MCP server
- **Architecture**: Clean, minimal single-file approach following
established MCP server patterns
- **Dependencies**: Minimal footprint using only `@modelcontextprotocol/sdk`
and `axios`
- **Deployment**: Supports both NPX (`npx apache-druid-mcp`) and Docker
deployment methods
#### Core MCP Tools Provided
1. **`execute_sql_query`** - Execute SQL queries against Druid datasources
with proper error handling
2. **`list_datasources`** - Retrieve all available datasources with metadata
3. **`get_datasource_metadata`** - Get detailed schema and configuration for
specific datasources
4. **`test_connection`** - Verify connectivity and authentication with Druid
broker
#### Key Features
- **Secure Configuration**: Environment-based configuration for URLs,
credentials, and timeouts
- **Error Handling**: Comprehensive error handling with informative messages
for AI agents
- **Type Safety**: Full TypeScript implementation with proper type
definitions
- **Production Ready**: Docker support, proper logging, and configuration
validation
- **Documentation**: Complete deployment guides for both NPX and Docker
methods
#### Design Decisions
- **Single-file Architecture**: Following the pattern of successful MCP
servers (like Google Maps MCP), consolidated handlers into `index.ts` for
simplicity and maintainability
- **Environment Configuration**: Uses standard environment variables
(`DRUID_URL`, `DRUID_USERNAME`, etc.) for flexible deployment
- **Stdio Transport**: Uses MCP's standard stdio transport for seamless
integration with AI clients
- **Minimal Dependencies**: Only essential dependencies to reduce attack
surface and maintenance burden
#### Use Cases
- **AI-Powered Analytics**: Enable AI agents to query Druid data for
insights and reporting
- **Automated Monitoring**: AI tools can monitor datasource health and
performance
- **Dynamic Dashboards**: AI applications can build dynamic visualizations
based on Druid data
- **Data Discovery**: AI agents can explore and understand available
datasets in Druid
#### Release note
Added Apache Druid MCP (Model Context Protocol) Server that enables AI
agents and applications to interact with Druid through a standardized
interface. The MCP server provides tools for executing SQL queries, listing
datasources, retrieving metadata, and testing connections. Available via NPX
(`npx apache-druid-mcp`) and Docker deployment methods. This integration makes
Druid more accessible to AI-powered applications and workflows.
<hr>
##### Key added files in this PR
* `mcp-server/src/index.ts` - Main MCP server implementation with all tools
and handlers
* `mcp-server/src/druid-client.ts` - Druid API client with proper error
handling
* `mcp-server/package.json` - NPX-ready package configuration
* `mcp-server/Dockerfile` - Multi-stage Docker build for production
deployment
* `mcp-server/docker-compose.yml` - Docker Compose configuration
* `mcp-server/DEPLOY.md` - Comprehensive deployment documentation
* `mcp-server/README.md` - Project overview and quick start guide
<hr>
This PR has:
- [x] been self-reviewed.
- [x] added documentation for new or modified features or behaviors.
- [x] a release note entry in the PR description.
- [x] added comments explaining the "why" and the intent of the code
wherever would not be obvious for an unfamiliar reader.
- [x] added unit tests or modified existing tests to cover new code paths,
ensuring the threshold for code coverage is met.
- [x] been tested in a test Druid cluster.
**Note**: This MCP server is a standalone TypeScript application that
communicates with Druid via HTTP APIs. It doesn't modify core Druid
functionality but provides a new integration point for AI applications. The
server has been thoroughly tested with both NPX and Docker deployment methods.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
