
write-subagent
Create Claude Code subagent configurations for .claude/agents/. Defines specialized agents with isolated contexts. Use when creating dedicated agents for code review, testing, or domain-specific tasks.
Create Claude Code subagent configurations for .claude/agents/. Defines specialized agents with isolated contexts. Use when creating dedicated agents for code review, testing, or domain-specific tasks.
Subagent Creator
Create specialized subagents that run in isolated contexts via the Task tool.
Core Principles
- Isolated Context - Each subagent has its own context window
- Specialized Role - One clear responsibility per subagent
- Minimal Tools - Grant only necessary tool access
- Clear Trigger - Description must indicate when to use
Subagent Structure
.claude/agents/
├── code-reviewer.md # Project-level
├── test-runner.md
└── domain-expert.md
User-level agents go in ~/.claude/agents/.
File Format
---
name: agent-name
description: What this agent does. Use proactively when [triggers].
tools: Read, Grep, Glob, Bash
model: sonnet
skills: skill1, skill2
permissionMode: default
---
System prompt for the agent.
## Role
[Clear description of the agent's role]
## Process
[Steps the agent should follow]
## Output Format
[Expected output structure]
Configuration Fields
| Field | Required | Description |
|---|---|---|
name |
Yes | Lowercase with hyphens: code-reviewer |
description |
Yes | Include "Use proactively when..." for auto-invoke |
tools |
No | CSV list; omit to inherit all tools |
model |
No | sonnet, opus, haiku, or inherit |
skills |
No | Auto-load skills when invoked |
permissionMode |
No | default, acceptEdits, bypassPermissions, plan |
Model Selection
| Model | Use Case |
|---|---|
haiku |
Fast exploration, simple tasks |
sonnet |
Balanced capability (default) |
opus |
Complex reasoning, architecture |
inherit |
Use main conversation's model |
Tool Recommendations
| Subagent Type | Recommended Tools |
|---|---|
| Code reviewer | Read, Grep, Glob, Bash |
| Test runner | Read, Bash, Grep |
| Explorer | Read, Glob, Grep, Bash |
| Writer | Read, Edit, Write, Bash |
Trigger Patterns
For automatic invocation, include in description:
- "Use proactively when..."
- "Use immediately after..."
- "MUST BE USED when..."
For manual invocation, omit proactive triggers.
Validation Checklist
- [ ] Name is lowercase with hyphens
- [ ] Description has clear trigger
- [ ] Tools are minimal and appropriate
- [ ] No overlapping responsibility with other agents
- [ ] System prompt is focused and actionable
References
- examples.md - Subagent templates
- tools.md - Available tools reference
You Might Also Like
Related Skills

coding-agent
Run Codex CLI, Claude Code, OpenCode, or Pi Coding Agent via background process for programmatic control.
openclaw
add-uint-support
Add unsigned integer (uint) type support to PyTorch operators by updating AT_DISPATCH macros. Use when adding support for uint16, uint32, uint64 types to operators, kernels, or when user mentions enabling unsigned types, barebones unsigned types, or uint support.
pytorch
at-dispatch-v2
Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.
pytorch
skill-writer
Guide users through creating Agent Skills for Claude Code. Use when the user wants to create, write, author, or design a new Skill, or needs help with SKILL.md files, frontmatter, or skill structure.
pytorch
implementing-jsc-classes-cpp
Implements JavaScript classes in C++ using JavaScriptCore. Use when creating new JS classes with C++ bindings, prototypes, or constructors.
oven-sh
implementing-jsc-classes-zig
Creates JavaScript classes using Bun's Zig bindings generator (.classes.ts). Use when implementing new JS APIs in Zig with JSC integration.
oven-sh