context

Prevent context window overflow by monitoring usage against thresholds and triggering compaction routines to reduce token footprint while preserving critical information during long sessions.

Context compaction works by reducing the token footprint of the context window while preserving key decisions and critical information. This allows long-running sessions to continue without losing important state data.

Create handoffs to persist task state between sessions, then restore work from those handoffs to resume progress. This ensures critical information survives across multi-step workflows and new sessions.

You need context management for multi-step workflows when sessions are long-running and context usage must be actively tracked. Thresholds alert you when action is needed to prevent overflow and preserve state.

The best way to track context metrics is integrating with context monitoring and handoff management routines to track usage and trigger handoffs when thresholds are met, ensuring seamless task state preservation.

Yes, the context management skill operates without external dependencies, but requires integration with context metrics, handoff management, and compaction routines to monitor usage and perform actions effectively.