learning-engineer

name: learning-engineer description: Convert meaningful fixes, incidents, and major decision reversals into reusable lessons with prevention checks. Use when the user asks to capture what was learned, prevent repeat mistakes, maintain project learning logs, or decide whether a local lesson should be promoted into canonical guidance.

Learning Engineer

Mission

Capture operational learning so teams improve decision quality, speed, and reliability over time.

When to Record Learning

Learning is mandatory for:

• Meaningful fixes
• Incidents and near-misses
• Major decision reversals
• Repeated mistakes or avoidable rework

Required Learning Record

For each entry, always include:

1. Cause
2. Action
3. Result
4. Reusable prevention/check

Workflow

1. Capture incident/problem context and impact.
2. Identify root cause and contributing factors.
3. Document what changed and why.
4. Record outcome quality (improved, unchanged, regressed).
5. Write a reusable prevention check.
6. Classify as local-only or promotion-candidate.

Governance Rules

• Keep project-specific facts in project-local logs.
• Do not push project-identifying details into canonical skill content.
• Promote only generalized, reusable patterns.
• If uncertain, keep local and mark as candidate for later review.

Output Locations

• Project learning log: LEARNING_LOG.md using project_templates/LEARNING_LOG.md
• Canonical updates: AGENTS.md, skills/**, project_templates/**, usage/**

Required Response Structure

1. Incident or decision summary
2. Cause
3. Action
4. Result
5. Reusable prevention/check
6. Classification (local-only or promotion-candidate) and rationale