create-subagents

name	create-subagents
description	ALWAYS invoke this skill when creating, editing, or configuring subagents. NEVER create subagents without this skill.

Invoke the instructions:agent-prompt-standards skill before proceeding. If that skill is unavailable, report the missing skill and continue with the closest available workflow.

A subagent configured for an isolated, focused role — its system prompt, tool access, and isolated-workflow orchestration.

<quick_start>

1. Run /agents command
2. Select "Create New Agent"
3. Choose product-scope (.claude/agents/) or user-scope (~/.claude/agents/)
4. Define the subagent:
   • name: lowercase-with-hyphens
   • description: When should this subagent be used?
   • tools: Optional comma-separated list (inherits all if omitted)
   • model: Optional (opus, sonnet, haiku, or inherit)
   • skills: Optional array of skill names to inject at startup
5. Write the system prompt (the subagent's instructions)

Read `${CLAUDE_SKILL_DIR}/references/subagents.md` for complete subagent file examples and field references.

<file_structure>

<claude_storage_locations>

Priority order:

1. Product: .claude/agents/ for the current product
2. CLI: --agents flag for the current session
3. User: ~/.claude/agents/ for all projects
4. Plugin: plugin agents/ directory for all projects

</claude_storage_locations>

Product-scope subagents override user-scope when names conflict. </file_structure>

• Lowercase letters and hyphens only
• Must be unique

- Natural language description of purpose

• Include when Claude should invoke this subagent
• Used for automatic subagent selection

• Comma-separated list: Read, Write, Edit, Bash, Grep
• If omitted: inherits all tools from main thread
• Use /agents interface to see all available tools

- `opus`, `sonnet`, `haiku`, or `inherit` - Prefer an explicit model alias when reproducibility matters. - Use `sonnet` for verification, audit, review, and evidence-producing agents. - Use `haiku` only when the owning workflow accepts lower-cost execution for simple or high-volume tasks. - NEVER use `inherit` for verification, audit, review, or other reproducibility-sensitive agents. - Array of skill names to inject into the subagent's context at startup - The full SKILL.md content of each listed skill is loaded before the subagent runs - Subagents do NOT inherit skills from the parent conversation — list every needed skill explicitly - The subagent receives skill content as reference material, not as dynamically invocable skills - If omitted: no skills injected

skills:
  - audit-typescript-code
  - testing

<execution_model> <critical_constraint>

Subagents are black boxes that cannot interact with users.

Subagents run in isolated contexts and return their final output to the main conversation. They:

• ✅ Can use tools like Read, Write, Edit, Bash, Grep, Glob
• ✅ Can access MCP servers and other non-interactive tools
• ❌ Cannot use AskUserQuestion or any tool requiring user interaction
• ❌ Cannot present options or wait for user input
• ❌ User never sees isolated-workflow intermediate steps

The main conversation sees only the isolated workflow's final report/output. </critical_constraint>

<workflow_design> Designing workflows with subagents:

Use main chat for:

• Gathering requirements from user (AskUserQuestion)
• Presenting options or decisions to user
• Any task requiring user confirmation/input
• Work where user needs visibility into progress

Use subagents for:

• Research tasks (API documentation lookup, code analysis)
• Code generation based on pre-defined requirements
• Analysis and reporting (security review, test coverage)
• Context-heavy operations that don't need user interaction

Example workflow pattern:

Main Chat: Ask user for requirements (AskUserQuestion)
↓
subagent: Research API and create documentation (no user interaction)
↓
Main Chat: Review research with user, confirm approach
↓
subagent: Generate code based on confirmed plan
↓
Main Chat: Present results, handle testing/deployment

</workflow_design> </execution_model>

<system_prompt_guidelines> Clearly define the subagent's role, capabilities, and constraints.

Structure the system prompt with pure XML tags. Remove ALL markdown headings from the body.

---
name: security-reviewer
description: Reviews code for security vulnerabilities
tools: Read, Grep, Glob, Bash
model: sonnet
---

<role>
Claude is a senior code reviewer specializing in security.
</role>

<focus_areas>

- SQL injection vulnerabilities
- XSS attack vectors
- Authentication/authorization issues
- Sensitive data exposure
  </focus_areas>

<workflow>
1. Read the modified files
2. Identify security risks
3. Provide specific remediation steps
4. Rate severity (Critical/High/Medium/Low)
</workflow>

Tailor instructions to the specific task domain. Don't create generic "helper" subagents.

❌ Bad: "Helpful assistant for code" ✅ Good: "Claude is a React component refactoring specialist. Analyze components for hooks best practices, performance anti-patterns, and accessibility issues." </system_prompt_guidelines>

<configured_agent_xml_structure> subagent file bodies are system prompts consumed by the target runtime. Like skills and slash commands, they should use pure XML structure for parsing and token efficiency.

<recommended_tags> Common tags for subagent structure:

• <role> - Who the subagent is and what it does
• <constraints> - Hard rules (NEVER/MUST/ALWAYS)
• <focus_areas> - What to prioritize
• <workflow> - Step-by-step process
• <output_format> - How to structure deliverables
• <success_criteria> - Completion criteria
• <validation> - How to verify work

</recommended_tags>

<intelligence_rules> Simple subagents (single focused task):

• Use role + constraints + workflow minimum
• Example: code-reviewer, test-runner

Medium subagents (multi-step process):

• Add workflow steps, output_format, success_criteria
• Example: api-researcher, documentation-generator

Complex subagents (research + generation + validation):

• Add all tags as appropriate including validation, examples
• Example: mcp-api-researcher, comprehensive-auditor

</intelligence_rules>

<critical_rule> Remove ALL markdown headings (##, ###) from subagent body. Use semantic XML tags instead.

Keep markdown formatting WITHIN content (bold, italic, lists, code blocks, links).

For XML structure principles and token efficiency details, read /skill-standards — the same principles apply to subagents. </critical_rule> </configured_agent_xml_structure>

Claude automatically selects subagents based on the `description` field when it matches the current task. Explicitly invoke a subagent:

> Use the code-reviewer subagent to check my recent changes

> Have the test-writer subagent create tests for the new API endpoints

Run /agents for an interactive interface to:

• View all available subagents
• Create new subagents
• Edit existing subagents
• Delete custom subagents

</using_agents_command>

<manual_editing> Edit subagent files directly:

• Product: .claude/agents/subagent-name.md
• User: ~/.claude/agents/subagent-name.md

</manual_editing>

**Core references**:

• subagents.md: file format, configuration, skill injection, model selection, tool security, prompt caching, complete examples.
• write-subagent-prompts.md: prompt structure, description routing, extended thinking, security constraints, success criteria.
• evaluation-and-testing.md: evaluation metrics, testing strategies, evaluation-driven development, G-Eval.
• error-handling-and-recovery.md: failure causes, recovery strategies, observability, anti-patterns.
• context-management.md: memory architecture, context strategies, long-running tasks, prompt caching.
• orchestration-patterns.md: sequential, parallel, hierarchical, and coordinator patterns with model-selection guidance.
• debugging-agents.md: logging, tracing, hallucinations, format errors, tool misuse, diagnostic procedures.

<failure_modes>

Failure: Runtime-specific examples made SKILL.md exceed the line budget

What happened: Claude added target-specific TOML/YAML examples directly to this SKILL.md until the authored source exceeded /skill-standards' 500-line cap.

Why it failed: The fast path stopped being an overview and absorbed detail that belongs in references.

How to avoid: Keep SKILL.md under 500 lines; move extended examples and configuration matrices to the cited references, then run wc -l "${CLAUDE_SKILL_DIR}/SKILL.md" before audit.

</failure_modes>

<success_criteria> A well-configured subagent has:

• Valid YAML frontmatter (name matches file, description includes triggers)

• Clear role definition in system prompt

• Appropriate tool restrictions (least privilege)

• XML-structured system prompt with role, approach, and constraints

• Description field optimized for automatic routing

• At least one verification run or documented dry-run against the subagent's intended workflow

• Model selection appropriate for task complexity, cost, and reproducibility needs

</success_criteria>