Where are these agent definitions stored?

Agent definitions are stored as .md files in the .claude/agents/ directory for project-specific use or the ~/.claude/agents/ directory for global, cross-project access.

Can I limit what tools a sub-agent can use?

Yes. This skill helps you define specific tool lists to enforce security and focus, such as restricting an agent to read-only access or preventing it from executing bash commands.

Why is the description field so important in agent definitions?

Claude uses the description field to determine when to trigger a specific agent. This skill helps you include specific trigger terms and usage examples so the right agent is used for the right task.

What is a Claude Code sub-agent?

A sub-agent is a specialized AI assistant with a dedicated purpose, specific toolset, and unique system prompt that Claude can automatically invoke to handle domain-specific tasks within your terminal.

How do I test if my new agent is working?

You can test discovery by asking Claude a natural language question related to the agent's expertise; Claude should automatically mention it is invoking the specialized sub-agent to help.

Agent Definition Creator

Name: Agent Definition Creator
Author: andisab

byandisab

•

Herramientas para Desarrolladores

Standardizes the creation of specialized Claude Code sub-agents with optimized system prompts and tool configurations.

Acerca de

The Agent Definition Creation skill empowers developers to architect specialized, production-ready AI assistants within the Claude Code environment by enforcing Anthropic's best practices for agent scoping and discovery. It provides a structured framework for defining sub-agent responsibilities, selecting appropriate model tiers, and implementing precise tool restrictions, ensuring that custom assistants are invoked accurately through optimized natural language descriptions and concrete usage examples. This skill is essential for scaling complex development workflows where specialized domain expertise is required.

Características Principales

3 GitHub stars
Enforces the principle of least privilege through granular tool restriction strategies
Provides structured templates for actionable, role-based system prompts
Supports multi-model selection including Sonnet, Opus, and Haiku for task optimization
Generates YAML frontmatter following Anthropic's official sub-agent specifications
Optimizes agent descriptions with discovery-driven triggers and example dialogues

Casos de Uso

Designing a security auditor sub-agent to review code before production merges
Building a personal documentation assistant that spans multiple local development projects
Creating a project-specific database expert with read-only query analysis tools

Acerca de

Características Principales

3 GitHub stars
Enforces the principle of least privilege through granular tool restriction strategies
Provides structured templates for actionable, role-based system prompts
Supports multi-model selection including Sonnet, Opus, and Haiku for task optimization
Generates YAML frontmatter following Anthropic's official sub-agent specifications
Optimizes agent descriptions with discovery-driven triggers and example dialogues

Casos de Uso

Designing a security auditor sub-agent to review code before production merges
Building a personal documentation assistant that spans multiple local development projects
Creating a project-specific database expert with read-only query analysis tools