aria-skill-candidates
Skill0 starsChanged 2 months agoApache-2.0View on GitHub
Observer + Gatekeeper for the Aria Builders meta-loop. Scans Claude Code transcripts for recurring vendor-API patterns and ranks them by a 5-axis rubric (recurrence · YAML-shape · multi-step · public-value · sovereignty-cost) to decide which patterns should become new skills. Filters out vendors already covered by an existing skill. Use when the user says "what should I build next", "skill candidates", "where am I repeating myself", "scan transcripts", or weekly as a passive review.
In this repository17 other files it gives its agents.
- aria-skill-templateSkill
- aria-skill-testSkill
- call-scrubSkill
- cloudflare-dns-deploySkill
- doc-to-pdfSkill
- ffmpeg-audioSkill
- github-repo-deploySkill
- goal-to-planSkill
- headless-claudeSkill
- live-code-watchSkill
- premortemSkill
- screen-describeSkill
- session-cabinetSkill
- skill-manifest-genSkill
- sprint-scaffoldSkill
- surgical-patchSkill
- video-ingestSkill
The file
Files
aria-skill-candidates
The "should I build a skill for this?" question, answered automatically.
Usage
python3 aria_skill_candidates.py # scan all transcripts
python3 aria_skill_candidates.py --limit-files 20 # last 20 sessions only
python3 aria_skill_candidates.py --json # machine-readableThe 5-axis rubric
Each candidate scored 0-3 on:
- Recurrence — how often the pattern appears
- YAML-shape — how cleanly it maps to declarative reconcile
- Multi-step — manual-step count if unskilled
- Public-value — generalizes vs Aria-internal
- Sovereignty — touches doctrine/splat/voice (subtracted)
Score = R + Y + M + P − S. ≥10 = BUILD NOW · 5-9 = LOG PROBE · <5 = SKIP.
License
Apache 2.0
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.