Mirror Mcp

This server gives LLMs a way to ask themselves questions through a single reflect tool that uses MCP's sampling capabilities to generate responses. You pass in a question, optional context, and sampling parameters like temperature and max tokens, and it returns a self-assessment response. Useful when you want the model to validate its own reasoning, check confidence levels, or catch logical gaps before committing to an answer. You can also pass custom system and user prompts to steer the reflection style, like asking it to think as a critical reviewer or focus on specific weaknesses. It's a metacognitive loop built on top of the standard MCP sampling interface.