@alexxx-db
基本情報
This project implements a Model Context Protocol (MCP) server that exposes Databricks Genie API capabilities as tools
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"databricks-genie-mcp": {
"command": "python",
"args": [
"server.py"
]
}
}
}Start a new conversation in a Genie space.
Create a new message in an existing conversation.
Retrieve a message from a conversation.
Get SQL query results from a message attachment.
Execute SQL for a message query attachment.
Get details about a Genie space.
Initiate a full query result download.
Poll a message until it reaches a terminal state.
It is a Model Context Protocol (MCP) server that exposes Databricks Genie API capabilities as tools. It enables integration of Databricks' no-code AI/BI assistant with other applications, allowing natural language querying of Databricks data through a standardized interface.
Install Python 3.10+, clone the repository, navigate to genie_api/, install dependencies (pip install -r requirements.txt), and configure authentication via a .env file with Databricks host and credentials (PAT or OAuth). Run python server.py locally. Then configure an MCP client like Claude Desktop to launch the server using the absolute path to server.py and the correct working directory.
Python 3.10+, a Databricks workspace with Genie access and system tables enabled (if using them), Databricks Assistant enabled, CAN USE permission on a Pro or Serverless SQL warehouse, access to Unity Catalog data, and an MCP-compatible client (e.g., Claude Desktop).
Create a `.env` file in the `genie_api/` directory with either PAT credentials (`DATABRICKS_HOST`, `DATABRICKS_TOKEN`) or OAuth with service principal (`DATABRICKS_HOST`, `DATABRICKS_CLIENT_ID`, `DATABRICKS_CLIENT_SECRET`). Use OAuth for production and never hardcode credentials.
Install Claude Desktop, add a JSON entry to its `mcpServers` config in `claude_desktop_config.json` with a `command` (e.g., `python`), `args` (full path to `server.py`), and `workingDirectory` (full path to `genie_api/` directory). Restart Claude Desktop; tools should appear under the hammer icon.
Never hardcode credentials; use environment variables and OAuth with service principals for production. The server runs locally with your permissions, and clients must obtain user consent before executing tools. For broader deployment, work with security/DevOps to secure hosting and credential access.
Verify Databricks credentials in the `.env` file and required permissions. Ensure absolute paths are correct in the Claude Desktop config, and check Claude Desktop logs (`~/Library/Logs/Claude/` on macOS, `%APPDATA%\Claude\logs` on Windows) for `mcp*.log` files. Try running `python server.py` manually from the `genie_api` directory to test.
Multi-zone MCP server for GKE/EKS/AKS: Python tools from git, NeMo guardrails, SSO, autoscaling.
Agent-Native Terminal
MCP server for AI Agents — exposes Yocoolab feedback threads and design selections as tools
Modelglass turns "which model should I use?" into a lookup instead of a guess. Seven tools give your agent live pricing across image, language, video, and audio models. Sourced and versioned, not scraped once and forgott
Routebase keeps your API's design, docs, contract tests, mocks and monitoring in one living OpenAPI spec so they can't drift apart. Agents work on that same source of truth, under your team's permissions.
コメント