If your issue tracker has accumulated dozens of reports that smell like duplicates but nobody has the nerve to close them, this is the clustering pass you've been putting off. It reads every open issue plus comment threads, groups symptoms by shared architectural root cause, opens one plan-master issue per cluster with a design doc in `plans/`, and closes children with a standardized redirect. The end state is open issues equals open plans, one to one. After the initial reduction it handles triage for new bugs and bundles PRs that close entire clusters atomically with a test matrix that prevents reoccurrence. Works best on backlogs of twenty plus issues where you suspect five real problems hiding under thirty surface complaints. Ships the discipline to keep it that way.
npx -y skills add thedotmack/claude-mem --skill oh-my-issues --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
sickn33/antigravity-awesome-skills
kubesphere/kubesphere
supercent-io/skills-template