strategic-compact

Manual context compaction prevents AI context loss by prompting compaction at logical task intervals, preserving critical information across phases instead of relying on arbitrary auto-compaction.

Strategic context compaction is a workflow optimization technique that tracks tool calls and prompts manual context clearing at phase boundaries, like after planning, to maintain session continuity.

Setting up manual context compaction requires a shell environment for script execution and configuring user-defined hooks in settings.json to track tool calls and prompt compaction at a configurable threshold.

Manual context compaction differentiates from arbitrary auto-compaction by targeting phase-aware intervals, prompting you to clear bulky research notes while preserving implementation plans before writing code.

You should compact context in complex AI tasks after completing the planning phase but before starting implementation, ensuring your implementation plan remains accessible while clearing out bulky research notes.