context-rot-detection

Context Rot Detection is an MCP server designed to provide AI agents with real-time self-awareness of their cognitive state, specifically detecting and monitoring 'context rot'—a decline in performance as long-running agent context windows become saturated. It offers health scoring, model-specific degradation analysis, actionable recommendations, health history tracking, and supports a wide range of popular and HuggingFace LLMs. This tool is valuable for developers and organizations running persistent AI agents who want to maintain high agent reliability and performance over time.