Databricks Explorer MCP

Databricks Explorer MCP is a prototype Minimal Command Protocol (MCP) interface designed for deep research and interactive data exploration via ChatGPT. It allows metadata exploration and SQL execution in a Databricks environment using a simplified search/fetch mechanism.

⚠️ Note: This is a conceptual example designed for future integration with ChatGPT. While the structure is functional in a server-hosted context, ChatGPT cannot directly execute SQL or external fetch requests yet. This MCP serves as a foundation for such future capabilities.

---

💡 Purpose

This tool demonstrates how ChatGPT could eventually support real-time, SQL-driven research through simple commands. By building a unified abstraction for metadata discovery and SQL querying, it bridges the gap between conversational interfaces and data platforms like Databricks.

---

🚀 Features

• 🔍 Search Tool:

  • Discover catalogs, schemas, and tables via keyword search.
  • Detects SQL-like input and creates a placeholder query::<sql> ID.

• 💥 Fetch Tool:

  • Executes SQL queries using a fixed warehouse (only in external environments).
  • Returns Unity Catalog metadata for catalog/schema/table IDs.

• 🧠 SQL-aware Search:

  • Input beginning with sql: or SQL verbs (SELECT, INSERT, etc.) is interpreted as a query stub.

• 🌐 FastMCP Compatible:

  • Built on the FastMCP framework.
  • Supports SSE transport for interactive use.

---

⚙️ Environment Variables

Variable	Description
DATABRICKS_WORKSPACE_URL	Databricks workspace base URL
DATABRICKS_TOKEN	Personal access token
DATABRICKS_WAREHOUSE_ID	Warehouse ID for executing SQL
PORT	(Optional) Server port, default 8080
LOG_LEVEL	(Optional) Logging level, default DEBUG

---

🆔 ID Format Summary

Type	Format
Catalog	catalog::<catalog>
Schema	schema::<catalog>.<schema>
Table	table::<catalog>.<schema>.<table>
SQL	query::<SQL statement>

---

🧩 Architecture

FastMCP
│
├── search(query) → metadata or SQL stub
└── fetch(id)     → SQL result or metadata

---

⚠️ ChatGPT Limitations

• ChatGPT cannot currently execute live SQL or access external APIs.
• This MCP is intended as a proof-of-concept and backend logic must be hosted separately.
• Useful for simulating integrations and planning future assistant capabilities.

---

📦 Installation & Run

pip install fastmcp requests
python server.py

Then access via:

http://localhost:8080/sse

---

🧪 Development Setup

Create and Activate a Virtual Environment

python3 -m venv .venv
source .venv/bin/activate

Install Dependencies

pip install -r requirements.txt

Run with MCP Inspector (Optional)

You can test the MCP interface using the Model Context Protocol Inspector:

npx @modelcontextprotocol/inspector@latest