Mirror

Mirror MCP provides AI assistants with self-reflection capabilities through a 'reflect' tool that enables recursive questioning and introspection. Built by Toby Padilla using the Model Context Protocol SDK, it uses MCP sampling to generate thoughtful responses to self-directed questions, with support for custom system and user prompts to create specialized reflection perspectives like coaching, critical thinking, or strategic analysis. The implementation includes robust fallback mechanisms when sampling fails and supports configurable parameters for temperature and token limits, making it valuable for AI systems that need to engage in metacognitive processes, self-assessment, or structured introspection to improve their reasoning and decision-making capabilities.

Source

Repository: https://github.com/toby/mirror-mcp