Skip to content
Back to skills

Learn

ASecurity

Manually add new learnings (patterns, preferences, rules) to memory. Use when you want to remember specific information.

  • 8 stars
  • 0 votes
  • 1 copy
  • 11 views
  • Added February 7, 2026
ai-agentstypescriptgobash

Security analysis

A100/100

Scanned February 12, 2026

npx -y skills add seokan-jeong/team-shinchan --skill learn --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Learn?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Learn
[![Security: A β€” Skills Directory](https://www.skillsdirectory.com/api/skills/seokan-jeong-learn/badge)](https://www.skillsdirectory.com/skills/seokan-jeong-learn)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: team-shinchan:learn
description: Manually add new learnings (patterns, preferences, rules) to memory. Use when you want to remember specific information.
user-invocable: true
---

# Learn Skill

**Manually teach Team-Shinchan something to remember.**

## Usage

```bash
/team-shinchan:learn "Always use TypeScript strict mode in this project"
/team-shinchan:learn "User prefers functional components over class components"
/team-shinchan:learn "This project uses pnpm, not npm"
```

## Process

### Step 1: Receive Learning Input

User provides something to remember.

### Step 2: Auto-Categorize

| Keywords in Input | Category |
|-------------------|----------|
| prefer, like, want | `preference` |
| pattern, approach, way | `pattern` |
| always, never, rule | `convention` |
| error, mistake, bug, avoid | `mistake` |
| decided, architecture, design | `decision` |
| use, using, project uses | `convention` |
| other | `insight` |

### Step 3: Save to Learning File

**File**: `.team-shinchan/learnings.md`

**Create file if it doesn't exist**, then append:

```markdown
## [YYYY-MM-DD HH:MM] {category}: {title}

**Context**: Manually taught by user
**Learning**: {full learning content}
**Confidence**: high
**Tags**: #{auto-generated tags}

---
```

### Step 4: Confirm

```
🧠 [Learn] Saved to memory:

Category: {category}
Learning: "{content}"
Confidence: high

πŸ“ Location: .team-shinchan/learnings.md
πŸ’‘ This will be applied in future sessions.
```

## Examples

**Input**: `/learn "Use Zustand for state management, not Redux"`

**Output**:
```
🧠 [Learn] Saved to memory:

Category: preference
Learning: "Use Zustand for state management, not Redux"
Confidence: high

πŸ“ Location: .team-shinchan/learnings.md
πŸ’‘ This will be applied in future sessions.
```

**Saved to file**:
```markdown
## [2024-02-03 16:00] preference: Use Zustand over Redux

**Context**: Manually taught by user
**Learning**: Use Zustand for state management, not Redux
**Confidence**: high
**Tags**: #state-management #zustand #redux #preference

---
```

## Important

- Learnings are **project-specific** (saved in `.team-shinchan/`)
- High confidence because user explicitly taught it
- Will be loaded at every session start
- Can be removed with `/team-shinchan:forget`

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.