Python Testing MCP Server

I've built a couple of MCPs and what jumps out about this repo is the clean split into three MCP tools—unit, fuzz, and coverage—so an LLM can sequence tests just like any other commands. By keeping the AI layer thin and the orchestration declarative, you could swap in Hypothesis or mutation-testing backends without rewriting the workflow. It's a solid pattern for anyone grafting AI onto an existing unittest stack.