learning-systems

To quantify implicit feedback from swarm task outcomes, use this Skill to translate task results into actionable signals without explicit user input, directly improving decomposition quality.

Confidence decay is a mechanism that tracks and adjusts pattern relevance over time to prevent stale guidance. It manages pattern maturity by automatically weighting and deprecating outdated patterns across tasks.

To detect and handle anti-patterns in a learning loop, the Skill automatically identifies patterns that degrade over time and manages their evolution, applying persistence to pattern states and errors.

To record swarm task outcomes, set up a sample swarm task and use the swarm_record_outcome function, which updates scores, decay, and maturity transitions within the pattern set.

Yes, this Skill operates with no external dependencies and is specifically designed for the opencode-swarm-plugin environment to debug learning loops and tune implicit feedback.