What is the difference between an agent and a skill in Claude Code?

Agents are specialized, isolated assistants meant for delegated tasks with their own context, whereas skills are 'always-on' capabilities that provide context-aware assistance automatically.

How do I determine which tool permissions to give an agent?

Follow the principle of least privilege: start with 'Read, Grep, Glob' for research, add 'Write, Edit' for code modification, and only add 'Bash' if system operations are strictly necessary.

Can a Claude Code subagent spawn another subagent?

No, subagents have a critical limitation where they cannot use the Task tool to spawn additional subagents. Orchestration of multiple agents should be handled by a skill.

Where are my custom agent files stored?

Agent files are stored as Markdown files in .claude/agents/ for project-level access, or ~/.claude/agents/ for user-level global access.

Claude Agent Architect

Name: Claude Agent Architect
Author: aiskillstore

byaiskillstore

•

114

Productivity & Workflow

Facilitates the creation, modification, and standardization of specialized Claude Code subagents with optimal configurations.

About

This skill empowers developers to architect sophisticated Claude Code environments by providing expert guidance on subagent creation and management. It helps users decide between agents, skills, and commands, ensures strict compliance with YAML schemas, enforces naming conventions, and optimizes tool permissions for enhanced security. By automating agent file generation and offering proactive validation, it streamlines the process of building modular, specialized AI assistants tailored to specific project needs within the Claude ecosystem.

Key Features

Strategic tool selection and security permission auditing
Architectural guidance for subagent isolation and context management
Automated agent file generation with validated YAML frontmatter
Visual branding via standardized terminal color palette recommendations
Integrated validation for agent naming, models, and capabilities
114 GitHub stars

Use Cases

Standardizing agent definitions and metadata across a development team
Designing a modular multi-agent architecture for complex coding projects
Converting repetitive manual workflows into specialized, isolated subagents

About

Key Features

Strategic tool selection and security permission auditing
Architectural guidance for subagent isolation and context management
Automated agent file generation with validated YAML frontmatter
Visual branding via standardized terminal color palette recommendations
Integrated validation for agent naming, models, and capabilities
114 GitHub stars

Use Cases

Standardizing agent definitions and metadata across a development team
Designing a modular multi-agent architecture for complex coding projects
Converting repetitive manual workflows into specialized, isolated subagents