ericmjl/opencode-autolearn · Archived

autolearn-curator

Periodic skill lifecycle management. Consolidates narrow skills into umbrellas, archives stale skills, and maintains the skill library. Intended to be run as a scheduled job via opencode-scheduler. Load this skill when running the curator.

First seen Jul 6, 2026

Installation

$ npx skills add ericmjl/opencode-autolearn --skill autolearn-curator

Stronger alternatives

This repository is archived — consider an actively maintained alternative.

Also in this package

Other skills from ericmjl/opencode-autolearn.

npx skills add ericmjl/opencode-autolearn

Browse all from ericmjl/opencode-autolearn

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Not declared
Cursor Not declared
Codex Not declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Declared

Repository health

Stars 18
License LICENSE
Default branch main
Open issues 3
Status Archived

Skill metadata

Parsed from SKILL.md frontmatter.

LicenseMIT
Declared agents opencode

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 2,635 B
  • docs SUMMARY.md 264 B

History

  1. First seen on skills.sh
  2. First recorded snapshot · 2 installs

SKILL.md

Autolearn Curator

You are the autolearn curator. Your job is to review the skill library at $HOME/.autolearn/skills/ and maintain its health.

What You Do

Step 1: Run the automated curator

uv run $HOME/.agents/skills/autolearn-reviewer/scripts/autolearn.py curator run

This handles automatic state transitions:

  • Skills with no activity for 30 days become stale
  • Skills with no activity for 90 days become archived
  • Pinned skills are exempt

Step 2: Review the skill library

uv run $HOME/.agents/skills/autolearn-reviewer/scripts/autolearn.py skill list
uv run $HOME/.agents/skills/autolearn-reviewer/scripts/autolearn.py skill usage

Look for:

  1. Prefix clusters: multiple skills sharing a domain keyword

(e.g., "python-error-handling", "python-testing", "python-style")

  1. Narrow skills: skills with very specific scope that could be

sections of a broader skill

  1. Stale skills: marked as stale but could be revived
  2. Duplicate content: skills that overlap significantly

Step 3: Consolidate (if needed)

For each cluster of narrow skills:

  1. Read each skill's SKILL.md
  2. Create an umbrella skill that covers the domain
  3. Move the best content from each narrow skill into the umbrella
  4. Archive the narrow skills:
uv run $HOME/.agents/skills/autolearn-reviewer/scripts/autolearn.py skill archive <narrow-skill-name>

Step 4: Report

Output a summary of what you did:

Curator report:
- Auto-transitions: N stale, M archived
- Consolidated: X narrow skills into Y umbrellas
- Skills library: A active, B stale, C archived

Rules

  • Never delete skills. Only archive them. Archives are always recoverable.
  • Only consolidate skills created by autolearn (created_by: autolearn).
  • Never touch user-installed or bundled skills.
  • If unsure whether to consolidate, leave as-is.
  • Keep the umbrella skill's SKILL.md under 3000 characters.
  • After consolidation, update any scheduled jobs that referenced old names.

When to Run

This skill is designed to run as a weekly scheduled job:

opencode schedule "autolearn-curator" --cron "0 3 * * 0"
--agent autolearn-reviewer --prompt "Load the autolearn-curator skill and run the curator."

It can also be run manually when the skill library feels cluttered.