context-degradation

Context degradation is the decline in AI agent reliability during long-running conversations caused by issues like lost-in-middle, context poisoning, and distraction. This skill diagnoses these patterns and provides quantitative health scoring to restore performance.

You can detect lost-in-middle and context poisoning by analyzing multi-turn agent sessions with this skill's modular scripts. It identifies degradation patterns, generates risk assessments, and provides actionable recommendations to mitigate context loss.

Yes, you can monitor AI context health in production using this skill, which relies on numpy for quantitative scoring. It evaluates high-context workloads and multi-turn sessions to pinpoint degradation and suggest remediation steps.

The best way to diagnose agent performance degradation during debugging is using structured detection scripts that score context health. This skill isolates confusion and distraction patterns in multi-turn sessions to provide actionable remediation guidance.

You should apply context degradation detection when context grows, performance degrades, or failures occur in long-running conversations, debugging sessions, or production deployments. It is essential for maintaining reliability in high-context workloads.