š¢ **New: signalk-mcp-server š**
š¢ New: signalk-mcp-server š
I have released signalk-mcp-server, a Node.js implementation of the Model Context Protocol (MCP) tailored for Signal K. It turns your Signal K instance into a first-class tool server for LLMs, exposing boat state in structured contexts.
š What it does:
ā¢ Implements MCP resources for querying, subscribing, and updating Signal K data
ā¢ Hosts a WebSocket/STDIO endpoint speaking MCPāno hand-rolled parsing
ā¢ Translates Signal K paths into MCP-friendly responses (e.g., vessel position, sensor data, system health)
ā¢ Enables integration with smart agents (Home Assistant, LangChain, Claude, etc.) š” Why this matters: Rather than polling Signal K JSON via REST or parsing streams, you can launch an LLM or agent, point it at this serverās MCP endpoint, and itāll "understand" boat contextāincluding histories, subscriptions, and live updatesāusing standard MCP semantics. This makes building marine-focused AI workflows, semantic automations, voice assistants, or monitoring dashboards far simpler. š Open source and ready for contributions: https://github.com/tonybentley/signalk-mcp-server
š Published on npm:
ā¢ Hosts a WebSocket/STDIO endpoint speaking MCPāno hand-rolled parsing
ā¢ Translates Signal K paths into MCP-friendly responses (e.g., vessel position, sensor data, system health)
ā¢ Enables integration with smart agents (Home Assistant, LangChain, Claude, etc.) š” Why this matters: Rather than polling Signal K JSON via REST or parsing streams, you can launch an LLM or agent, point it at this serverās MCP endpoint, and itāll "understand" boat contextāincluding histories, subscriptions, and live updatesāusing standard MCP semantics. This makes building marine-focused AI workflows, semantic automations, voice assistants, or monitoring dashboards far simpler. š Open source and ready for contributions: https://github.com/tonybentley/signalk-mcp-server
š Published on npm:
signalk-mcp-server https://www.npmjs.com/package/signalk-mcp-server
Contributions, feedback, and issues welcome!1 Reply
This looks very cool! I'm excited to play with it later this week when I have some free time.