What model should I use for my Claude Code agent?

It is recommended to use 'inherit' as the model setting. This allows the agent to use the same model as the parent process unless the task requires specific capabilities of Sonnet, Opus, or Haiku.

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

Agents are designed for autonomous, multi-step subprocesses that handle complex work independently, while commands are intended for specific, user-initiated actions.

Can I limit the tools an agent has access to?

Yes, you can use the optional 'tools' field in the YAML frontmatter to define an array of allowed tool names, following the principle of least privilege.

How do I ensure my agent triggers at the right time?

Effective triggering relies on a well-written description field in the agent's frontmatter, containing specific 'Use this agent when' conditions and 2-4 concrete usage examples.

Claude Code Agent Development

Name: Claude Code Agent Development
Author: lusya123

bylusya123

0•

Developer Tools

Guides developers through creating, configuring, and optimizing autonomous agents within the Claude Code ecosystem.

The Agent Development skill is a comprehensive resource for building autonomous subprocesses that handle complex, multi-step tasks within Claude Code. It provides standardized templates for YAML frontmatter, best practices for system prompt design, and precise rules for defining triggering conditions. Whether you are building agents for code reviews, automated testing, or API documentation, this skill ensures your agents are correctly namespaced, utilize the appropriate models, and follow the principle of least privilege through tool restriction.

Key Features

01Structured system prompt patterns for autonomous agent behavior

02Validation rules for agent identifiers, colors, and tool sets

030 GitHub stars

04Standardized YAML frontmatter templates for agent metadata

05Best practices for crafting effective triggering conditions and examples

06Workflows for both manual and AI-assisted agent generation

Use Cases

01Restricting agent capabilities to specific tools for enhanced security and focus

02Optimizing agent triggering logic to ensure proactive and reactive task handling

03Building specialized sub-agents for niche tasks like security auditing or UI analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lusya123/cloudcode agent-development

For use in Claude.ai and ChatGPT

Download Skill