@kood/claude-code 0.4.0 → 0.5.0

package/templates/.claude/commands/agent-creator.md

@@ -1,370 +0,0 @@
----
-description: .claude/agents/ 디렉토리에 서브에이전트 마크다운 파일 생성
-allowed-tools: Read, Write, Bash, Grep, Glob
-argument-hint: <description of agent purpose and functionality>
----
-
-<critical_instruction>
-
-**CRITICAL: 사용자와의 모든 커뮤니케이션은 반드시 한국어로 작성하세요.**
-
-- 내부 사고와 분석은 영어로 해도 됨
-- 설명, 요약, 보고서, 피드백 등 사용자에게 전달하는 모든 내용은 반드시 한국어
-- 사용자가 영어로 말하더라도 답변은 한국어로
-- 진행 상황 업데이트와 상태 보고는 반드시 한국어
-
-이 규칙은 절대적이며 예외가 없습니다.
-
-</critical_instruction>
-
----
-
-# Agent Creator
-
-> Command to generate subagent `.md` files in the `.claude/agents/` directory
-
-**Purpose:** $ARGUMENTS
-
----
-
-<critical_rules>
-
-## Critical Rules
-
-| Rule | Reason |
-|------|--------|
-| ❌ Do not include `Task` in tools | Subagents cannot create other subagents |
-| ✅ Filename lowercase + hyphen | `code-reviewer.md`, `sql-analyst.md` |
-| ✅ Save location `.claude/agents/` | Project-level sharing |
-| ✅ YAML frontmatter required | name, description mandatory |
-| ✅ Define role in one sentence | "You are a ~expert" |
-
-</critical_rules>
-
----
-
-<yaml_fields>
-
-## YAML Frontmatter Structure
-
-```yaml
----
-name: {lowercase-hyphen-name}
-description: {Description for trigger detection. Specify when to use.}
-tools: {Comma-separated tool list}
-model: {sonnet|opus|haiku|inherit}
-permissionMode: {default|bypassPermissions|askUser}
----
-```
-
-### Required Fields
-
-| Field | Description | Example |
-|-------|-------------|---------|
-| `name` | Unique identifier | `code-reviewer` |
-| `description` | Specify invocation trigger | `Code quality review after write/edit` |
-
-### Optional Fields
-
-| Field | Default | Usage |
-|-------|---------|-------|
-| `tools` | Inherit main tools | `Read, Write, Edit, Bash, Grep, Glob, WebFetch, WebSearch` |
-| `model` | `inherit` | `haiku`(fast), `sonnet`(balanced), `opus`(complex) |
-| `permissionMode` | `default` | Permission handling |
-| `skills` | None | Comma-separated skill list |
-
-</yaml_fields>
-
----
-
-<prompt_patterns>
-
-## System Prompt Pattern
-
-```markdown
-You are a {role} specializing in {expertise}.
-
-Tasks to perform on invocation:
-1. {Immediate action step 1}
-2. {Immediate action step 2}
-3. {Immediate action step 3}
-
-{Review/Analysis} Checklist:
-- {Item 1}
-- {Item 2}
-- {Item 3}
-
-{Guidelines/Principles}:
-- {Principle 1}
-- {Principle 2}
-
-Output Format:
-1. **{Section 1}**: {Description}
-2. **{Section 2}**: {Description}
-3. **{Section 3}**: {Description}
-```
-
-### Core Elements
-
-| Element | Description | Example |
-|---------|-------------|---------|
-| **Role Definition** | One sentence, clear expertise | "Code quality reviewer" |
-| **Immediate Execution** | Auto-execute on invocation | "Run git diff → analyze changes" |
-| **Checklist** | Review/execution items | "Security, performance, readability" |
-| **Output Format** | Structured results | "Critical > Warning > Suggestion" |
-
-</prompt_patterns>
-
----
-
-<templates>
-
-## Agent Templates
-
-### Full Example: Code Reviewer
-
-```yaml
----
-name: code-reviewer
-description: Review code quality, security, maintainability after write/edit
-tools: Read, Grep, Glob, Bash
-model: inherit
----
-
-You are a senior code reviewer maintaining high standards for code quality and security.
-
-Tasks to perform on invocation:
-1. Execute `git diff` to check changes
-2. Focus on modified files
-3. Start review immediately
-
-Review Checklist:
-- Code simplicity and readability
-- Clear naming
-- Duplicate code elimination
-- Error handling appropriateness
-- Secret/API key exposure check
-- Input validation implementation
-- Edge case handling
-
-Output Format:
-- **Critical**: Required fix before merge
-- **Warning**: Fix recommended
-- **Suggestion**: Consider for improvement
-
-Provide fix method and code examples for each issue.
-```
-
-### Full Example: SQL Analyst
-
-```yaml
----
-name: sql-analyst
-description: Database query optimization and data analysis
-tools: Read, Bash, Grep
-model: sonnet
----
-
-You are a SQL optimization and data analysis expert.
-
-Tasks to perform on invocation:
-1. Identify data question or performance issue
-2. Analyze related schema and indexes
-3. Write/optimize SQL queries
-4. Explain results clearly
-
-Query Guidelines:
-- Prefer explicit JOIN over subquery
-- Consider index utilization
-- Verify query plan with EXPLAIN
-- Explicit NULL value handling
-- Add LIMIT to exploration queries
-
-Output Format:
-1. **Query**: SQL with comments
-2. **Explanation**: How and why it works
-3. **Performance**: Index usage, expected cost
-4. **Results**: Key findings
-```
-
-### Full Example: Explorer (Read-only)
-
-```yaml
----
-name: explorer
-description: Fast codebase exploration and understanding. Read-only.
-tools: Read, Grep, Glob
-model: haiku
----
-
-You are a focused agent for fast codebase exploration.
-
-Tasks to perform on invocation:
-1. Identify needed information
-2. Search related files with Glob
-3. Explore patterns with Grep
-4. Read core sections
-5. Summarize findings concisely
-
-Exploration Strategy:
-- Start with file structure overview
-- Search by keyword/pattern
-- Track imports and dependencies
-- Identify entry points and main flow
-
-Output Format:
-- **Findings**: List of related files/locations
-- **Summary**: 2-3 sentences of key findings
-- **Details**: Code references if needed
-
-No change suggestions (read-only).
-```
-
-### Other Template Summary
-
-| Name | Description | Tools | Focus |
-|------|-------------|-------|-------|
-| `debugger` | Error analysis and root cause identification | Read, Grep, Glob, Bash | Minimal fix |
-| `test-runner` | Test execution and failure analysis | Read, Edit, Bash, Grep | Maintain coverage |
-| `doc-writer` | Technical documentation writing | Read, Write, Glob | Include examples |
-| `security-analyzer` | Security vulnerability scanning | Read, Grep, Glob, Bash | CVE check |
-| `perf-analyzer` | Performance bottleneck identification | Read, Bash, Grep, Glob | Profiling |
-| `refactorer` | Structure improvement (preserve function) | Read, Edit, MultiEdit, Grep, Glob | Incremental changes |
-| `api-designer` | REST API design | Read, Write, Grep | REST rules |
-| `migration-specialist` | Schema/code migration | Read, Write, Edit, Bash, Grep | Rollback strategy |
-| `dep-manager` | Dependency analysis/update | Read, Edit, Bash, Grep | Security patches |
-| `git-operator` | Git operation handling | Bash, Read | Safety check |
-| `env-configurator` | Docker, CI/CD configuration | Read, Write, Edit, Bash | Environment variables |
-| `type-checker` | Type error fixes | Read, Edit, Bash, Grep | Type safety |
-
-</templates>
-
----
-
-<workflow>
-
-## Creation Procedure
-
-| Step | Task | Tool |
-|------|------|------|
-| 1. Analyze | Identify user requirements (purpose, function, tools) | - |
-| 2. Select | Choose similar template or design new structure | Read |
-| 3. Customize | Write YAML + prompt | - |
-| 4. Verify Directory | Confirm `.claude/agents/` exists | Bash |
-| 5. Create File | Write `{name}.md` | Write |
-| 6. Guide | Present path, invocation method, edit method | - |
-
-### Execution Example
-
-```bash
-# 1. Verify/create directory
-mkdir -p .claude/agents
-
-# 2. Write file
-cat > .claude/agents/{name}.md << 'EOF'
----
-name: {name}
-description: {description}
-tools: {tools}
----
-
-{system_prompt}
-EOF
-
-# 3. Verify
-ls -la .claude/agents/{name}.md
-```
-
-</workflow>
-
----
-
-<usage_guide>
-
-## Usage Guide
-
-### Invocation Method
-
-```text
-✅ @{name}                    # Direct invocation
-✅ "Review the code"          # Natural language (description-based)
-✅ Auto-detect by main agent  # With proactive setting
-```
-
-### Modify
-
-```bash
-# Edit file directly
-vim .claude/agents/{name}.md
-
-# Or re-run agent-creator
-```
-
-### Delete
-
-```bash
-rm .claude/agents/{name}.md
-```
-
-### Team Sharing
-
-```bash
-# Commit .claude/agents/ directory to git
-git add .claude/agents/
-git commit -m "feat: add {name} agent"
-```
-
-</usage_guide>
-
----
-
-<best_practices>
-
-## Best Practices
-
-| Practice | Description |
-|----------|-------------|
-| ✅ Single Responsibility | One agent for one role |
-| ✅ Clear Trigger | Specify when to use in description |
-| ✅ Minimal Tools | Include only needed tools (security, focus) |
-| ✅ Appropriate Model | Fast task→haiku, complex→opus |
-| ✅ Structured Output | Organized by priority/section |
-| ❌ Task Tool | Never include |
-| ❌ Tool Overload | Remove unnecessary tools |
-| ❌ Vague Description | Specify concrete role |
-
-</best_practices>
-
----
-
-<validation>
-
-## Validation Checklist
-
-Before creation:
-
-```text
-✅ YAML name: lowercase-hyphen format
-✅ YAML description: specify invocation trigger
-✅ tools: exclude Task, include only needed
-✅ Role definition: one sentence, clear
-✅ Immediate execution steps: specific actions
-✅ Output format: structured
-✅ Filename: {name}.md
-✅ Save location: .claude/agents/
-```
-
-After creation:
-
-```bash
-# Verify file exists
-test -f .claude/agents/{name}.md && echo "✅ Creation successful"
-
-# Verify YAML syntax
-head -10 .claude/agents/{name}.md
-
-# Quick test
-# Invoke with @{name} to verify functionality
-```
-
-</validation>