From memento-skills
Use when the user wants to remove low-value learned skills, clean up unused skills, or reduce the learned skill library size
How this skill is triggered — by the user, by Claude, or both
Slash command
/memento-skills:memento-pruneThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Remove learned skills that are no longer useful. Never auto-deletes — always confirms with user.
Remove learned skills that are no longer useful. Never auto-deletes — always confirms with user.
Read metrics.json (global and project). Flag skills matching ANY:
utility < utility_threshold AND usage_count >= min_samples_for_judgment (proven low value)last_used older than prune_after_days_unused days from today (abandoned)Show each candidate with reasoning:
Prune candidates:
1. learned-bad-pattern (global)
Utility: 0.20 (1/5) — below threshold 0.40
Last used: 2026-01-15 — 69 days ago
Reason: low utility + unused
2. learned-old-approach (project)
Utility: — (0/0) — never used
Last used: 2026-01-20 — 64 days ago
Reason: abandoned (never used, 64 days old)
Ask: "Delete these skills? You can choose individually (e.g., 'just 1') or all."
For each confirmed deletion:
learned-{name}/ skill directoryEven if called programmatically, always show the list and wait for confirmation.
npx claudepluginhub tmdgusya/momento --plugin memento-skillsIdentifies and removes orphaned, unreferenced, or redundant skill content after quarterly maintenance or linter flags. Inverse of /skill-update.
Interviews the user, scans coding-agent log files, and recommends unused K-skills for removal. Supports Claude Code, Codex, OpenCode, OpenClaw, and Hermes Agent.
Scans Claude Code config (skills, memory, hooks, permissions, MCP servers, caches) for stale or redundant items and walks the user through confirm-each-deletion cleanup.