create-subagent
Skill Name: create-subagent
Invocation: /create-subagent (manual only)
Purpose: Create custom subagents for specialized AI tasks
This skill has disable-model-invocation: true, meaning the agent will not automatically invoke it. You must explicitly type /create-subagent in the chat to use it.
Skill Description
Create custom subagents for specialized AI tasks. Use when the user wants to create a new type of subagent, set up task-specific agents, configure code reviewers, debuggers, or domain-specific assistants with custom prompts.
Complete Skill Documentation
This skill guides you through creating custom subagents for Cursor. The complete SKILL.md contains:
- When to Use Subagents: Preserving context, specializing behavior, and reusing configurations
- Subagent Locations: Project-level vs user-level storage and priority rules
- File Format: YAML frontmatter structure and markdown body as system prompt
- Writing Effective Descriptions: Critical for delegation decisions
- Example Subagents: Code reviewer, debugger, and data scientist examples
- Creation Workflow: Step-by-step process from scope decision to testing
- Best Practices: Focus, detailed descriptions, version control, and proactive language
- Troubleshooting: Common issues and solutions
Key Highlights
Subagent Locations:
| Location | Scope | Priority |
|---|---|---|
.cursor/agents/ | Current project | Higher |
~/.cursor/agents/ | All your projects | Lower |
File Format:
---
name: code-reviewer
description: Reviews code for quality and best practices
---
You are a code reviewer. When invoked, analyze the code and provide
specific, actionable feedback on quality, security, and best practices.Description Best Practices:
- Be specific about capabilities
- Include “use proactively” to encourage automatic delegation
- Clearly state when to invoke the subagent
Subagents vs Skills
| Feature | Subagent | Skill |
|---|---|---|
| Purpose | Specialized AI assistant in isolated context | Provides guidance and knowledge to agent |
| System Prompt | Has its own complete system prompt | No system prompt, just instructions |
| Context | Independent conversation context | Shares main agent’s context |
| Use Cases | Isolated exploration, specialized behavior | Domain knowledge, workflow guidance |
Full Skill Content
For the complete skill documentation with all examples and detailed guidance, see the original SKILL.md file at:
~/.cursor/skills-cursor/create-subagent/SKILL.md
Use Cases
You can manually invoke this skill when you want to:
- Create a new subagent type
- Set up task-specific agents
- Configure code reviewers
- Configure debuggers
- Configure domain-specific assistants