Helping LLMs and Agents write good code is all about context

I found myself often copying and pasting logs/errors from my dev server 

So I made a lightweight MCP server that wraps any terminal command, and exposes the logs. Perfect to connect to Claude Code, Cursor, Windsurf, etc

Introducing: server-watch-mcp

https://github.com/BrennerSpear/server-watch-mcp

Once you've set it up with a few commands, you can add update your package.json, no changes to your workflow needed

"dev": "server-watch-mcp next dev --turbo"

Decentralized open source consciousness | Values-aligned communities | citizen of Black Rock City

software artist | building /outpaint | ex: @coinbase ex: bisontrails | seeker of the flow state | artivilla.com

I would guess that any system connected to MCP servers has a list of all the functions of it can call in its context and this one has “get logs” and “search logs”

how does it know it has to use that specific mcp tool and not another?