curate

Archives stale, abandoned, or failing skills recoverably after explicit user approval.

20|2|Updated May 16, 2026
One-click install
npx skills add https://github.com/reddb-io/red-skills --skill curate-reddb-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: curate
Source: https://github.com/reddb-io/red-skills/tree/main/plugins/dev/skills/engineering/curate
Command: npx skills add https://github.com/reddb-io/red-skills --skill curate-reddb-io

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Over time, projects accumulate skills that go stale, get loaded but never used, or fail repeatedly, cluttering the skill library and degrading agent behavior. This Skill identifies those Curatable-skill candidates and archives them recoverably, so cleanup never risks permanent data loss. ## Core Features & Use Cases - Category-grouped curation: Reads memory curate skills --json and groups candidates into stale, abandoned, frequently-failing, and archive categories with evidence for each. - Approval-gated archiving: Archives only skills the user explicitly names, moving them via atomic rename into .red/memory/skill-archive/ with a SHA-256 hash manifest recording the approving category. - Recoverable restore: Reverses any archive with /curate --restore <name>, verifying every restored file's hash against the manifest. - Background mode: With --background, files a single ready-for-human Issue listing candidates and never mutates any Skill file. - Use Case: After months of agent work, run /curate to review skills with no activity for 90 days, approve the ones to archive, and restore any of them later if needed. ## Quick Start Run /curate to review stale, abandoned, and frequently-failing skills grouped by category, then reply with the names you approve for recoverable archiving.

Frequently Asked Questions about curate

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I archive unused or stale skills safely?▼

Run /curate to list candidates grouped by stale, abandoned, frequently-failing, and archive categories, then approve specific names. Each archive uses atomic rename into .red/memory/skill-archive/ with a SHA-256 manifest, and reverses with /curate --restore <name>.

How do I restore an archived skill?▼

Run /curate --restore <name>. The engine reads the archive manifest, moves the skill back to its original path, and verifies every restored file's SHA-256 against the manifest before completing.

Can I run skill curation non-interactively in the background?▼

Yes, /curate --background files a single ready-for-human Issue listing candidates grouped by category. It performs zero filesystem mutations; a human later runs interactive /curate to approve and archive.

Why does memory curate check fail before curation starts?▼

The boot precondition requires graph mode with skill telemetry enabled. If it exits non-zero, the stderr message names the exact memory init --mode graph --skill-telemetry command needed to fix it.

Which skills are excluded from curation archiving?▼

Bundled skills with source_kind plugin or hub and pinned skills are filtered defensively in every category and never archived. Only Curatable skills, meaning user-owned or agent-created ones, are eligible.