cross-task-learning

Aggregating insights across tasks involves collecting reflections and structured observations from each completed task, then analyzing them to detect recurring patterns, bottlenecks, and best practices. This Skill applies deterministic updates to a centralized aggregate file, enabling you to surface cross-task trends that guide data-driven evolution.

Pattern detection uses similarity-based matching across aggregated reflections with threshold-driven actions to classify recurring behaviors as best practices or anti-patterns. The Skill processes structured reflections into a memory file, exposing patterns that repeat across diverse tasks for prioritization.

Yes. By aggregating reflections and identifying patterns across tasks, the Skill generates evolution proposals ranked by frequency and impact. These proposals surface bottlenecks and improvement opportunities discovered through cross-task analysis, enabling you to prioritize system improvements data-driven.

The Skill works within Reflector and Evolver contexts, consuming individual task reflections and writing deterministic updates to a shared aggregate.json memory file. It transforms scattered task learnings into a cohesive knowledge base that informs iterative evolution.

Reflections must follow a defined protocol to enable deterministic processing and similarity matching. The Skill expects consistent structured input—typically JSON reflections with task outcomes, observations, and learnings—to reliably detect patterns and update the aggregate knowledge store.