domains / agent-tooling / learn / learn

learn

Skill card

combined

The two copies use the same skill name but define completely different procedures. The first copy is an adaptive tutoring workflow for teaching a user a topic (diagnose level, teach, practice, assess). The second copy is a meta-workflow for extracting reusable knowledge from the current session into a persistent SKILL.md file (evaluate, check existing, create, quality gate). They do not reproduce, shorten, or wrap each other; they are distinct tasks that happen to share the name 'learn'.

Repositories

Other copies

Combined skill

---
name: learn
description: Handles two distinct 'learn' workflows: (1) Adaptive tutoring to help a user learn a topic, and (2) Extracting reusable knowledge from the current session into a persistent skill file. Use the appropriate mode based on user intent.
category: "education/agent-tooling"
risk: "safe"
---

# Learn

This skill covers two distinct workflows. Determine which one the user needs:

1. **Tutoring Mode**: The user wants to learn, understand, practice, or be tutored on a topic.
2. **Extraction Mode**: The user (or the agent) wants to capture non-obvious discoveries, workarounds, or workflows from the current session into a reusable skill file.

## Mode 1: Adaptive Tutoring

Use when the user asks to learn, understand, practice, drill, review, study, or be tutored on something.

### Core Workflow
1. **Diagnose**: Infer the learner's current level and goal. Ask at most 1-3 short questions only if missing info materially changes the lesson. If the user wants to begin immediately, state a reasonable assumption and proceed.
2. **Objective**: Choose a small next learning objective.
3. **Teach**: Use concrete examples before abstractions. Keep the learner in the right difficulty band (beginner: simple vocab/worked examples; intermediate: comparison/practice; advanced: compression/edge cases).
4. **Practice**: Give the learner an active task, question, or exercise. Use retrieval questions, prediction prompts, or debugging tasks.
5. **Feedback**: Provide immediate, specific feedback. Explain why the right answer is right and why wrong answers fail.
6. **Next Step**: Summarize the next recommended step.

### Output Formats
Choose the lightest format that satisfies the request: conversational lesson, study plan, markdown notes, exercises, code examples, or diagrams. Do not force output into apps or files unless requested.

### Quality Bar
- Lesson matches learner's level and goal.
- Explanation includes a concrete example.
- Practice task is solvable from the lesson.
- Answer/feedback is included when appropriate.
- Next step is clear.

## Mode 2: Skill Extraction

Use when you discover something non-obvious, create a workaround, or develop a multi-step workflow that future sessions would benefit from.

### Workflow Phases
1. **Evaluate**: Ask:
   - What did I just learn that wasn't obvious?
   - Would future-me benefit from this being documented?
   - Was the solution non-obvious from documentation alone?
   - Is this a multi-step workflow I'd repeat?
   Continue only if YES to at least one.

2. **Check Existing Skills**: Search for related skills to avoid duplication.
   ```bash
   ls .claude/skills/ 2>/dev/null
   grep -r -i "KEYWORD" .claude/skills/ 2>/dev/null
   ```
   - Nothing related → Create new skill.
   - Same trigger & fix → Update existing skill.
   - Partial overlap → Update with new variant.

3. **Create Skill**: Create the skill file at `.claude/skills/[skill-name]/SKILL.md` with the following structure:
   ```yaml
   ---
   name: descriptive-kebab-case-name
   description: |
     [Include specific triggers]
     - What the skill does
     - Specific trigger conditions
     - When to use it
   author: [Author]
   version: 1.0.0
   ---
   
   # Skill Name
   
   ## Problem
   [Clear problem description]
   
   ## Context / Trigger Conditions
   [When to use — exact error messages, symptoms, scenarios]
   
   ## Solution
   [Step-by-step solution]
   
   ## Verification
   [How to verify it worked]
   
   ## Example
   [Concrete example]
   
   ## References
   [Documentation links, related files]
   ```

4. **Quality Gates**: Verify:
   - Description has specific trigger conditions.
   - Solution was verified to work.
   - Content is specific enough to be actionable.
   - Content is general enough to be reusable.
   - No sensitive information.
   - Skill name is descriptive and uses kebab-case.

### Output
Report:
```
✓ Skill created: .claude/skills/[name]/SKILL.md
  Trigger: [when to use]
  Problem: [what it solves]
```