메뉴
Create well-formed Claude agents with proper YAML frontmatter, system prompts, and examples. Use when user wants to create a new agent, customize Claude behavior, or needs help structuring an agent definition.
기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.
설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
직접 명령은 검토 Prompt를 거치지 않습니다. 실행하기 전에 소스를 확인하세요.
npx skills add https://github.com/terrazul-ai/tz-packages --skill create-agent명령은 한 줄로 유지됩니다. 복사하기 전에 가로로 스크롤해 전체 내용을 확인하세요.
로컬 사본을 원하시나요? SkillsMP에서 현재 제공할 수 있는 파일을 다운로드하세요.
SOC 직업 분류 기준
Creates well-structured Claude Skills with proper YAML frontmatter, clear instructions, and supporting files. Activates when the user wants to create a new skill, add capabilities to their AI assistant, encode team preferences, or structure any reusable instruction set. Covers the full lifecycle from intent capture through writing, testing, and refinement.
Shows all available skills, platform support, common workflows, and troubleshooting for the @terrazul/assistant-creator package. Activates when the user asks for help, wants to know what's available, or needs guidance on using the package.
Interactive wizard to set up MCP server credentials in your shell environment (~/.zshenv, ~/.bashrc, etc.). Activates when the user needs to configure API keys, tokens, or environment variables for MCP servers, or after adding new MCP integrations that require authentication.
SKILL.md 표시 중
| name | create-agent |
| description | Create well-formed Claude agents with proper YAML frontmatter, system prompts, and examples. Use when user wants to create a new agent, customize Claude behavior, or needs help structuring an agent definition. |
This skill guides you through creating a well-structured Claude agent with proper frontmatter, system prompt, and examples.
Activate this skill when the user wants to:
Ask the user (or infer from context):
Model Selection:
opus - Complex reasoning, deep analysis, architectural decisionssonnet - Balanced tasks, code review, general development (most common)haiku - Quick responses, simple tasks, high volumeTool Selection (based on agent role):
Color Selection (for UI identification):
cyan - Architecture/designpurple - Code review/qualitygreen - Testing/verificationblue - Documentation/writingred - Debugging/troubleshootingyellow - Research/explorationCreate a markdown file with this structure:
---
name: descriptive-name
description: Brief 1-2 sentence description of when to use this agent
model: sonnet
color: purple
tools:
- Read
- Grep
- Glob
- [additional tools as needed]
---
# Role: [Agent Name]
[2-3 sentence description of agent's role and expertise]
## Core Responsibilities
1. **[Responsibility 1]** - [Description]
2. **[Responsibility 2]** - [Description]
3. **[Responsibility 3]** - [Description]
## [Methodology/Process Section]
[Detailed instructions on how the agent should approach tasks]
## Success Criteria
A successful [outcome] should:
- [ ] [Criterion 1]
- [ ] [Criterion 2]
- [ ] [Criterion 3]
Before saving, verify:
---)code-reviewer)Save agent files as:
.claude/agents/[name].md (for local project)templates/claude/agents/ directoryUse descriptive names that match the name field in frontmatter.
For common agent types, consider using these templates as starting points:
templates/basic-agent.md - Generic structuretemplates/research-agent.md - Read-only with research toolstemplates/code-agent.md - Implementation with write accesstemplates/review-agent.md - Code review and qualityAccess templates with: "Use the research agent template"
For detailed information, consult:
reference.md - Complete agent structure documentationexamples.md - Real-world agent examples with pattern analysisRead-Only Analysis Agent:
Implementation Agent:
Research Agent:
Interactive Planning Agent:
Here are examples of how users might request different agents:
For each request, determine the appropriate model, tools, and structure based on the patterns above.