Can subagents ask me questions while they work?

No, subagents operate as black boxes and cannot use the AskUserQuestion tool. They must be given all necessary context upfront and return a final report to the main conversation.

How do I trigger a specific subagent?

Claude can automatically select a subagent based on its description, or you can explicitly invoke one using natural language commands like 'Use the code-reviewer subagent'.

Where are my custom subagents stored?

Subagents can be stored at the project level in .claude/agents/ or at the user level in ~/.claude/agents/ for use across all your projects.

Why should I use XML tags in subagent prompts?

Claude parses XML structures more efficiently and accurately than markdown headings, leading to better instruction adherence and higher token efficiency.

What is a Claude Code subagent?

A subagent is a specialized instance of Claude that runs in an isolated context with a focused role and specific tool access, designed to complete tasks autonomously.

Subagent Architect

Name: Subagent Architect
Author: glittercowboy

byglittercowboy

•

770

Productivity & Workflow

Streamlines the creation and management of specialized Claude Code subagents for autonomous, focused task execution.

About

This skill provides expert guidance for architecting, building, and orchestrating Claude Code subagents and utilizing the Task tool. It enables developers to delegate complex, research-heavy, or repetitive tasks to isolated Claude instances that run autonomously without user interaction. By mastering subagent configuration—including granular tool access, model selection, and XML-structured system prompts—users can build sophisticated multi-agent workflows that significantly increase development velocity and focus.

Key Features

Expert guidance for writing optimized XML-structured system prompts
Project-level and global user-level agent configuration management
Multi-agent orchestration patterns for sequential and parallel workflows
Interactive subagent creation via the /agents command interface
Granular tool permission management for least-privilege security
770 GitHub stars

Use Cases

Setting up specialized agents for automated test generation and validation
Delegating deep API documentation research and synthesis to background agents
Creating autonomous code reviewers that analyze changes for security and quality

About

Key Features

Expert guidance for writing optimized XML-structured system prompts
Project-level and global user-level agent configuration management
Multi-agent orchestration patterns for sequential and parallel workflows
Interactive subagent creation via the /agents command interface
Granular tool permission management for least-privilege security
770 GitHub stars

Use Cases

Setting up specialized agents for automated test generation and validation
Delegating deep API documentation research and synthesis to background agents
Creating autonomous code reviewers that analyze changes for security and quality