context-engineering

Token optimization involves archiving stale information and loading only relevant context as needed. This Skill prunes irrelevant conversation history and manages your active context window, reducing token use by up to 84% while maintaining performance.

Context rot occurs when outdated or irrelevant information accumulates in an AI's active context over long sessions, consuming tokens and pushing out high-signal data. Curating context by removing stale material and loading task-specific guidelines prevents performance degradation.

Active context curation identifies stale information, archives it to a knowledge file, removes it from active context, and verifies quality. This Skill automates the workflow with triggers like conversation length, task switching, and tool outputs, plus manual commands like /context optimize.

Yes. This Skill loads project-specific guidelines from CLAUDE.md when a new task begins, ensuring relevant context is active and irrelevant details are archived, keeping the AI focused and efficient throughout complex operations.

Use context optimization before complex operations, after tool use, when switching tasks, or after major learnings. Automatic triggers fire based on conversation length thresholds, ensuring your active context stays high-signal without manual intervention.