context-optimize

Context compression reduces conversation drift by automatically summarizing and pruning long documents and tool results. It reclaims context headroom by applying self-editing techniques to remove stale inputs during planning, coding, reviews, and debugging.

Pruning stale tool results is best handled by triggering compress and prune commands. This applies self-editing context techniques using H2O and Context-1 principles with configurable heuristics to automatically extract key information and clean up extended AI sessions.

Context compression works by applying configurable heuristics and H2O plus Context-1 principles to summarize long inputs. It uses self-editing techniques to extract key information, automatically reducing conversation drift and reclaiming context headroom during multi-file projects.

Context pruning is applicable for debugging and code review workflows. It enables pre-cleaned context across multi-file projects and extended AI sessions, automatically compressing and summarizing long documents to improve responsiveness during these tasks.

You should compress context to reclaim headroom when long documents and tool results cause conversation drift in extended AI sessions. Running compress and prune operations extracts key information, preventing stale results from degrading responsiveness during multi-file projects.

No dependencies are required to run context optimization. The Skill operates independently, applying H2O and Context-1 principles with configurable heuristics to summarize and prune long documents across multi-file projects without external tool requirements.