Subagent Delegation & Workflow FAQs

Question 1

What types of specialized agents are available?

Accepted Answer

The skill includes pre-defined agents for code review, security auditing (OWASP), type design analysis, dead code hunting (loose-ends-hunter), and general codebase exploration.

Question 2

How does this skill integrate with Gemini?

Accepted Answer

The skill includes optional scripts to delegate to Gemini for tasks requiring massive context (up to 1M tokens), such as full-repo diff-based security or design reviews, while using Claude subagents for complex logic and user-journey analysis.

Question 3

Why should I delegate tasks to subagents instead of keeping them in one chat?

Accepted Answer

Aggressive delegation prevents context window 'overloading.' By spawning fresh subagents, you give each task a focused environment with specific instructions, leading to higher accuracy and more thorough analysis than a single overloaded session.

Question 4

Can I run multiple subagents at once?

Accepted Answer

Yes, the skill supports a Parallel Dispatch Pattern. You can launch multiple independent investigations (e.g., a security review and a mobile UI review) simultaneously within a single message block to save time.

Question 5

Can I create custom agents for my specific project?

Accepted Answer

Absolutely. You can define custom project agents in your .claude/agents/ directory to handle specific requirements like design system compliance or internal architectural patterns.

Subagent Delegation & Workflow

Subagent Delegation & Workflow

Key Features

Use Cases

Key Features

Use Cases