What is a Claude Code Agent?

An agent is an autonomous subprocess within a Claude Code plugin designed to handle complex, multi-step tasks independently using specific tools and specialized system prompts.

What are the naming conventions for agents?

Agent identifiers must be between 3-50 characters, use only lowercase letters, numbers, and hyphens, and must start and end with an alphanumeric character.

Which models can a Claude agent use?

Agents can be configured to use 'inherit' (to match the parent session), 'sonnet' for balanced performance, 'opus' for complex reasoning, or 'haiku' for fast, low-cost tasks.

How do I trigger an agent in Claude Code?

Agents are triggered through the 'description' field in their frontmatter. This field must include 'Use this agent when...' conditions followed by multiple XML-style example blocks showing the scenario, user request, and desired assistant response.

Claude Agent Development

Name: Claude Agent Development
Author: jawhnycooke

byjawhnycooke

0•

Productivity & Workflow

Architects and implements high-performance autonomous agents for Claude Code plugins using standardized frameworks and best practices.

The Agent Development skill provides a comprehensive framework for building autonomous sub-processes within the Claude Code ecosystem. It guides developers through the technical requirements of agent creation, including YAML frontmatter configuration, system prompt engineering, and the implementation of precise triggering conditions. By utilizing this skill, users can design specialized agents that handle complex, multi-step tasks independently, ensuring their Claude Code plugins are robust, predictable, and capable of sophisticated autonomous behavior.

Key Features

01Advanced system prompt engineering patterns and templates

02Model selection and tool-restriction configuration (least privilege)

03Standardized agent file structure and YAML frontmatter guidance

04Precise agent triggering logic using example-driven descriptions

050 GitHub stars

06Automated validation rules for identifiers and metadata

Use Cases

01Developing a 'Test Generator' agent that handles edge-case analysis independently

02Creating a specialized 'Security Auditor' subagent for autonomous code reviews

03Building documentation agents that monitor and update API references automatically

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add jawhnycooke/claude-plugins agent-development

For use in Claude.ai and ChatGPT

Download Skill