Sub-agent usage in Claude Code has evolved from manual orchestration to intelligent automation. This guide covers the two primary approaches to sub-agent utilization: manual sub-agents and custom agents.

Manual Sub-agents​

The original approach using the Task tool for explicit parallel processing:

Use 3 sub-agents to analyze these files:
1. Security analysis of auth.ts
2. Performance review of cache system
3. Type checking of utils.ts

• Benefits: Direct control, predictable behavior
• Drawbacks: Manual orchestration overhead, no tool/MCP selection control, shared system prompt inheritance, same model for all tasks

Custom Agents​

Specialized agents with isolated contexts, custom system prompts and tool selection that activate automatically (see Custom Agents for detailed configuration):

---
name: security-reviewer
description: Security analysis specialist for authentication and authorization code
tools: Read, Grep, Bash
model: opus
---

You are a security expert specializing in authentication vulnerabilities...

• Benefits: Automatic activation, isolated contexts, token efficiency
• Drawbacks: Setup overhead, configuration complexity

Decision Framework​

Choose the right sub-agent approach based on your specific needs and workflow requirements:

Use Manual Sub-agents When:

• Simple parallel operations: File reads, searches, basic analysis.
• One-off analysis: Multi-perspective reviews where you want to specify exactly which viewpoints to use.
• Quick turnaround needed: No setup overhead.
• You want explicit control: Direct orchestration of which sub-agents handle which tasks.
• Non-destructive work: Research, analysis, comparison matrices.

Use Custom Agents When:

• Specialized expertise needed repeatedly: Code review, security analysis, performance optimization that you do across multiple projects.
• Domain-specific work: UX review, SEO optimization, technical writing, accessibility audits.
• Role-specific tool access required: Security agents that only have access to Read and Grep tools, not file modification tools.
• Long-term reusability: Build once, use everywhere.
• You prefer automatic delegation: Let Claude intelligently route tasks to the right specialist based on context.
• Team standardization: Share the same agent configurations across your entire team.
• Cross-project deployment: Refined agents that work instantly in new codebases.

See Also: Task Agent Tools|Custom Agents|Agent Engineering|Sub-agent Tactics