qmcp Server

qmcp Server is an MCP server that provides Model Context Protocol integration for connecting to and executing queries on q and kdb+ databases. Designed primarily for quantitative developers, financial engineers, and data analysts working with high-frequency time-series data, it allows AI coding assistants like Claude to interact directly with live development q sessions. The server uses a single persistent connection model with sequential query execution to prevent resource conflicts and mirror standard developer workflows. It incorporates intelligent asynchronous query switching, enabling immediate responses for fast calculations while maintaining progress tracking for heavy analytics tasks. Additionally, the server provides AI-controlled query interruption via SIGINT signals to cancel runaway computations and infinite loops. Through integrated experimental tools, qmcp Server also offers bidirectional translation capabilities between vectorized Python-like syntax and native q code, facilitating collaborative code generation and debugging directly within supported MCP environments.

Category: Databases & Data Stores

Tags: finance, kdb+, q, time-series

Visit qmcp Server

How to install and configure qmcp Server

1. Ensure Python 3.8+ and a running q/kdb+ database instance are available (e.g., started with q -p 5001). 2. To install globally, install the package using pip: pip install qmcp. 3. Add the server to Claude CLI by running: claude mcp add qmcp qmcp (or run via uv: claude mcp add qmcp "uv run qmcp/server.py"). 4. For Claude Desktop, edit your configuration file and add the server definition under mcpServers: json { "mcpServers": { "qmcp": { "command": "qmcp" } } } 5. Restart your client, connect to your active q port via the connect_to_q tool, and issue queries.

What you can do with qmcp Server

• Querying financial time-series data and computing analytics directly on running kdb+ instances from Claude. * Interrupting long-running or runaway q queries using programmatic SIGINT signals equivalent to Ctrl+C. * Managing persistent development connections to local or remote q processes with custom async timeouts. * Translating experimental vectorized Python code snippets into q expressions using the Qython translation utilities. * Inspecting table schemas, metadata, and task execution statuses during interactive q application debugging sessions.

Key facts

• https://github.com/gabiteodoru/qmcp
• Databases & Data Stores
• finance, kdb+, q, time-series

Part of MCP Servers

What is qmcp Server?

qmcp Server is an open-source Model Context Protocol server that bridges AI assistants with q and kdb+ time-series database instances. It allows AI clients to establish persistent connections, execute database queries sequentially, manage query timeouts, and interrupt running tasks.

Which MCP clients work with qmcp Server?

The server works with Claude Desktop and the Claude CLI. While the core database querying and connection management tools work across MCP-compatible clients, certain experimental code translation features are specifically wired to work with Claude Code CLI.

Can qmcp Server interrupt runaway queries on Windows?

Query interruption via SIGINT is disabled when running directly on Windows outside WSL, or when the MCP server and the q process run across the Windows and WSL boundary. Windows users should run both components inside WSL.

What tools are included in qmcp Server?

The server includes production-ready tools for connecting to q instances, executing queries, configuring connection and async timeouts, checking async task status, and canceling queries. It also includes experimental tools for translating between q and Python-like syntax.