create-subagent
Guide for creating specialized Claude Code subagents with proper YAML frontmatter, focused descriptions, system prompts, and tool configurations. Use when the user wants to create a new subagent, custom agent, specialized AI assistant, or mentions creating/designing/building agents or subagents.
npx skills add majiayu000/claude-skill-registry --skill create-subagent-andy-austin-quipu --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Guides the user through creating a specialized Claude Code subagent, including purpose, location, model choice, tool access, system prompt structure, and testing steps. It provides a standardized Markdown file format with YAML frontmatter that defines name, description, model, and tools, and then a detailed system prompt layout for the subagent.
How it works
- Step-by-step Quick Start: understand purpose, choose location, select model, define tool access, write a system prompt, and test invocation.
- File Format: requires a Markdown file with YAML frontmatter containing fields: name, description, model (optional), tools (optional).
- Configuration Fields: required fields are name and description; model and tools are optional (tools inheritance or restriction).
- File Locations: supports project-level (.claude/agents/) and user-level (~/.claude/agents/) scopes, with plugin-level as an option.
- Description Best Practices: emphasizes clarity and explicit delegation triggers like "Use PROACTIVELY" or similar MUST BE USED cues.
- Model Selection Guide: Haiku for simple tasks, Sonnet for balanced work, Opus for advanced; Inherit to match main model.
- Tool Restrictions: examples show how to restrict or allow tools (Read, Write, Bash, etc.).
- System Prompt Structure: detailed sections for identity, purpose, capabilities, behavioral traits, knowledge base, response approach, and example interactions.
- Invocation Methods: automatic delegation (preferred) and explicit invocation commands.
- Creation Checklist and Testing: lists validation steps, restart guidance, and CLI testing examples.
When to use it
Use when the user wants to create a new subagent, a custom agent, or a specialized AI assistant, or explicitly mentions creating/designing/building agents or subagents.
What it can touch
The skill references tools such as Read, Write, Edit, Bash, Grep, Glob, WebFetch, Task, and others; it documents restricted vs. unrestricted tool configurations and how to declare them in YAML frontmatter.
Caveats
- License: MIT
- Declared tools: claude-code
- File location guidance and frontmatter validity are required by the workflow; ensure YAML frontmatter is valid and name constraints (lowercase, numbers, hyphens) are followed.
# Create Subagent Guide This skill helps you create specialized Claude Code subagents following official best practices and standards. Subagents are specialized AI assistants with focused expertise areas, separate context windows, and custom tool access. ## Quick Start When creating a new subagent, follow this workflow: 1. **Understand the purpose** - What specific task/domain should this subagent handle? 2. **Choose location** - Project-level (.claude/agents/) or user-level (~/.claude/agents/)? 3. **Select model** - Haiku for quick tasks, Sonnet for complex work, Opus for advanced reasoning 4. **Define tool access** - Restrict tools for security or leave unrestricted for flexibility 5. **Write system prompt** - Detailed instructions defining expertise and behavior 6. **Test invocation** - Verify automatic delegation and explicit invocation work ## File Format Every subagent is a Markdown file with YAML frontmatter: ```markdown --- name: your-subagent-name description: When and why this subagent should be invoked model: sonnet # Optional: sonnet, opus, haiku, or inherit tools: Read, Write, Bash, Grep # Optional: comma-separated list --- Your subagent's detailed system prompt goes h
- Quick Start
- File Format
- Configuration Fields
- File Locations
- Description Best Practices
- Model Selection Guide
- Tool Restrictions
- System Prompt Structure
- 1. Core Identity
- 2. Expert Purpose
- 3. Capabilities
- 4. Behavioral Traits
- 5. Knowledge Base
- 6. Response Approach
claude --agents '{
Add to version control
git add .claude/agents/your-agent.md
git commit -m "Add specialized agent for X"
git push
Project-level
Or user-levelWhat does the create-subagent skill do?
Guide for creating specialized Claude Code subagents with proper YAML frontmatter, focused descriptions, system prompts, and tool configurations. Use when the user wants to create a new subagent, custom agent, specialized AI assistant, or mentions creating/designing/building agents or subagents.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill create-subagent-andy-austin-quipu --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.