Rate and sync AI skills — find the best, skip the garbage.

Problem •
Quick Start •
Tools •
Commands •
Scoring

---

The Problem

You've got 400+ skills from Spawner, community repos, and your own custom ones. You sync them all to Claude, Cursor, Codex... and now:

- 5 skills claim "authentication" — which one does your AI use?
- Half are garbage — incomplete, no validations, zero sharp edges
- Contradictions everywhere — one skill says "use Prisma", another says "use Drizzle"

Blind sync tools make this worse. They dump everything and hope for the best.

The Solution

Genaura Sync rates every skill (0-100) before syncing. You see exactly which skills are worth using, which overlap, and which to delete.

``bash
npx genaura-sync rate
`

`
✔ Rated 437 skills

Quality Summary
───────────────────────────────────────
⚡ Excellent (80-100): 23 skills
✓ Good (60-79): 161 skills
○ Mediocre (40-59): 100 skills
✗ Poor (0-39): 153 skills ← these are polluting your context

Overlaps Detected
───────────────────────────────────────
"authentication" claimed by:
→ Backend Engineering (61) ← USE THIS ONE
→ Security Patterns (47)
→ Auth Helper (23) ← DELETE
`

Quick Start

`bash


Install globally

npm install -g genaura-sync

Or run directly

npx genaura-sync rate # See what you've got
npx genaura-sync sync # Sync the good stuff
npx genaura-sync clean # List garbage to remove
`

Supported Tools

Syncs to 8 AI coding tools simultaneously:

| Tool | Path |
|------|------|
| Claude | ~/.claude/skills |
| Cursor | ~/.cursor/skills |
| Codex | ~/.codex/skills |
| Copilot | ~/.copilot/skills |
| Gemini | ~/.gemini/skills |
| OpenCode | ~/.opencode/skills |
| Windsurf | ~/.windsurf/skills |
| Antigravity | ~/.antigravity/skills |

Skills are symlinked from your source (e.g., ~/.spawner/skills), so updates propagate automatically.

Commands

`bash


Rate & analyze

gsync rate # Scan and rate all skills
gsync rate -v # Verbose with score breakdown
gsync rate --json # Output as JSON for scripting

Sync (the main event)

gsync sync # Rate then sync to all targets
gsync sync --min-score 60 # Only sync skills scoring 60+
gsync sync --dry-run # Preview without changes
gsync sync --no-backup # Skip backup step

Find the best

gsync best "database" # Best skill for a use case
gsync compare next react vue # Side-by-side comparison

Maintain

gsync clean # List poor skills with rm commands
gsync status # Show configured paths
gsync targets # List sync targets
`

Scoring Rubric

Each skill is scored 0-100 across 4 categories:

| Category | Points | What It Measures |
|----------|--------|------------------|
| Identity | 25 | Role, expertise, triggers, ownership domains |
| Sharp Edges | 25 | Pitfalls documented, detection patterns, solutions |
| Validations | 25 | Rules with patterns, anti-patterns documented |
| Collaboration | 25 | Delegation rules, handoff triggers, compatible skills |

$3

| Tier | Score | Action |
|------|-------|--------|
| ⚡ Excellent | 80-100 | Use everywhere |
| ✓ Good | 60-79 | Use, maybe improve |
| ○ Mediocre | 40-59 | Use with caution |
| ✗ Poor | 0-39 | Delete with gsync clean |

How It Works

`
┌─────────────────────────────────────────────────────────────┐
│ 1. SCAN ~/.spawner/skills, ~/.claude/skills, etc. │
├─────────────────────────────────────────────────────────────┤
│ 2. SCORE Rate each skill 0-100 │
├─────────────────────────────────────────────────────────────┤
│ 3. DETECT Find overlaps, contradictions, conflicts │
├─────────────────────────────────────────────────────────────┤
│ 4. RECOMMEND "Use backend (61), skip auth-helper (23)" │
├─────────────────────────────────────────────────────────────┤
│ 5. SYNC Symlink to Claude, Cursor, Codex, etc. │
└─────────────────────────────────────────────────────────────┘
`

Real Example

`bash
$ gsync sync --min-score 60

╔═══════════════════════════════════════╗
║ ⚡ GENAURA SYNC ║
║ Rate skills. Use the best. ║
╚═══════════════════════════════════════╝

✔ Rated 437 skills from ~/.spawner/skills

Quality Summary
───────────────────────────────────────
⚡ Excellent: 23 ✓ Good: 161 ○ Mediocre: 100 ✗ Poor: 153

Syncing 184 skills (score ≥ 60), skipping 253

Overlaps Found
───────────────────────────────────────
"authorization" → Backend Engineering (61) vs Security (47)
"testing" → QA Engineering (61) vs Test Strategist (39)

Contradictions
───────────────────────────────────────
backend ↔ api-designer: Both claim rate-limiting, error-handling

Syncing
───────────────────────────────────────
✓ claude: 184 linked, 253 skipped
✓ cursor: 184 linked, 253 skipped
✓ codex: 184 linked, 253 skipped
✓ gemini: 184 linked, 253 skipped

⚡ Sync complete! Backup at ~/.config/genaura-sync/backups/
`

Why This Exists

I had 400+ skills synced everywhere. My AI kept using the wrong one — a half-baked "auth-helper" instead of the comprehensive "Backend Engineering" skill.

Existing sync tools (skillshare, etc.) just copy everything. No quality check. No overlap detection. No recommendations.

Genaura Sync fixes this. Rate first, sync smart.

Contributing

`bash
git clone https://github.com/genaura/genaura-sync
cd genaura-sync
npm install
npm run dev -- rate # Test locally
``

PRs welcome for:
- New sync targets
- Scoring improvements
- Better overlap detection

Links

- npm
- GitHub
- Genaura — AWS SaaS Control Panel

License

MIT © Genaura