What problem does it solve? Improvement work in a project tends to pile up: issues get filed but never prioritized, low-priority items starve forever, and no one steps back to find the root cause behind recurring symptoms. This Skill runs a self-improvement loop that periodically reviews the whole project, files root-cause improvement issues, grooms the backlog, and implements the top item autonomously. ## Core Features & Use Cases - Discovery mode: Launches parallel review agents across N project-defined perspectives (pipeline health, repository hygiene, artifact quality, codebase, strategy) to identify meta root causes and file deduplicated improvement issues with priority and story-point labels. - Grooming mode: Aggregates open improvement issues via a triage script, closes clear duplicates and stale items, creates Epic tracking issues, and runs a weekly refinement gate that forces every stagnant issue into one of four exits (work on it, drop it, bundle it, defer it) with circuit breakers against infinite deferral. - Consumption mode: Daily digestion of up to 5 ready issues sorted by priority with tie-breaking rules, hold circuit breakers, and automatic retrospective handoff after merges. - Use Case: A team maintaining an active GitHub repository schedules this Skill so that every week stagnant issues are refined, every day the top improvement is implemented and merged via PR, and every month the backlog is groomed into Epics. ## Quick Start Ask the AI to run a cross-cutting review of the project and file improvement issues for the root causes it finds.
