amnafarzy/ravie/pattern-learner
Claude Code and agent skill by amnafarzy/ravie
- community
Checking that the SKILL.md file is still in the repository.
Use this skill when repeated corrections, a significant feature, a slow debugging session, or a weekly review create evidence for reusable patterns and skill or CLAUDE.md updates. Do not use for one-off annoyances, immediate implementation, or unverified patterns without evidence.
Install
npx vskill@latest install amnafarzy/ravie/pattern-learnerSKILL.md
The same SKILL.md was published earlier as amnafarzy/ravie/pattern-learner.
pattern-learner
Weekly review: extract reusable patterns from the week's work, identify repeated friction, and propose improvements to skills and CLAUDE.md.
Purpose
The system that makes the system better. After each week of real use, review what happened and feed improvements back into skills, rules, and project context.
When to use this
- End of week (Friday or Monday morning)
- After shipping a significant feature
- After a debugging session that took too long
- After noticing the same correction 3+ times
When NOT to use this
- Do not use for one-off annoyances, immediate implementation, or unverified patterns without evidence.
Process
1. Pull the week's data
- Linear: issues closed, issues created, blockers encountered
- GitHub: PRs merged, review comments, failed CI
- Notion: decisions logged, runbooks updated
- VPS/automation: job failures, new automations
2. Identify patterns
Look for:
- Repeated corrections — same thing fixed 3+ times → update a skill or CLAUDE.md
- Repeated friction — same slow step every PR → automate or simplify
- Repeated bugs — same bug class recurring → add a rule or hook
- Unused skills — invoked but didn't help → improve or remove
- Missing skills — did something manually that should be a skill → create one
3. Score pattern confidence
- LOW — seen once. Note it, don't act yet.
- MEDIUM — seen 2-3 times. Draft the improvement.
- HIGH — seen 4+ times. Apply the improvement now.
4. Propose changes
For each HIGH confidence pattern:
- Which file to update (SKILL.md, CLAUDE.md, rules, hooks)
- What specific change to make
- Why (cite the evidence from this week)
5. Apply approved changes
After human approval, commit improvements to the ravie repo.
Output format
Hard rules
- Never skip the weekly review — this is what makes the system improve
- Never act on LOW confidence patterns — wait for repetition
- Never change skills based on theory — only change based on observed friction
- Always cite specific evidence for proposed changes
Connects to
skill-creator— for creating new skills from HIGH confidence patternsworkflow-evaluator— for evaluating whether existing skills are workingnotion-brain— for storing review outputlinear-operator— for pulling closed issuesgithub-operator— for pulling merged PRs
Common failure modes
Acting on one-off annoyance — Something annoyed you once, you propose a new skill. Wait for HIGH confidence (4+ occurrences) before changing the system. One-time friction often doesn't recur.
Proposing skill edits without evidence — "Maybe issue-to-pr should also do X." Stop. What specifically went wrong? Cite the session, the correction, the friction. Without evidence, the change is guessing.
Summarizing activity without converting to repo changes — "This week we shipped 8 PRs and closed 12 issues." Useful for retrospectives, useless for system improvement. Pattern-learner's output must be specific changes to specific files, with evidence.
Updating CLAUDE.md instead of skills — The fix goes in the most general place, bloating project context. Domain-specific patterns belong in the relevant skill or rule, not CLAUDE.md.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingarchive/skills/pattern-learner/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/amnafarzy/ravie/pattern-learner)