Files
everything-claude-code/.opencode/commands/skill-create.md

118 lines
2.0 KiB
Markdown
Raw Normal View History

---
description: Generate skills from git history analysis
agent: build
---
# Skill Create Command
Analyze git history to generate Claude Code skills: $ARGUMENTS
## Your Task
1. **Analyze commits** - Pattern recognition from history
2. **Extract patterns** - Common practices and conventions
3. **Generate SKILL.md** - Structured skill documentation
4. **Create instincts** - For continuous-learning-v2
## Analysis Process
### Step 1: Gather Commit Data
```bash
# Recent commits
git log --oneline -100
# Commits by file type
git log --name-only --pretty=format: | sort | uniq -c | sort -rn
# Most changed files
git log --pretty=format: --name-only | sort | uniq -c | sort -rn | head -20
```
### Step 2: Identify Patterns
**Commit Message Patterns**:
- Common prefixes (feat, fix, refactor)
- Naming conventions
- Co-author patterns
**Code Patterns**:
- File structure conventions
- Import organization
- Error handling approaches
**Review Patterns**:
- Common review feedback
- Recurring fix types
- Quality gates
### Step 3: Generate SKILL.md
```markdown
# [Skill Name]
## Overview
[What this skill teaches]
## Patterns
### Pattern 1: [Name]
- When to use
- Implementation
- Example
### Pattern 2: [Name]
- When to use
- Implementation
- Example
## Best Practices
1. [Practice 1]
2. [Practice 2]
3. [Practice 3]
## Common Mistakes
1. [Mistake 1] - How to avoid
2. [Mistake 2] - How to avoid
## Examples
### Good Example
```[language]
// Code example
```
### Anti-pattern
```[language]
// What not to do
```
```
### Step 4: Generate Instincts
For continuous-learning-v2:
```json
{
"instincts": [
{
"trigger": "[situation]",
"action": "[response]",
"confidence": 0.8,
"source": "git-history-analysis"
}
]
}
```
## Output
Creates:
- `skills/[name]/SKILL.md` - Skill documentation
- `skills/[name]/instincts.json` - Instinct collection
---
**TIP**: Run `/skill-create --instincts` to also generate instincts for continuous learning.