When should I use an Agent instead of a Command?

Use Agents for focused, background analysis requiring domain expertise where the user doesn't need interactive control. Use Commands for tasks requiring direct user interaction or step-by-step guidance.

What is a Claude Code Agent?

Agents are autonomous specialized workers within Claude Code that perform focused tasks independently, such as code reviews or performance analysis, returning results once their specialized work is complete.

Can agents coordinate with other agents?

Yes, using the Meta Agent pattern and the 'Task' tool, you can create orchestrator agents that spawn and manage sub-agents for multi-faceted validation or complex workflows.

What is the purpose of agent color-coding?

Colors provide visual identification in the UI: blue for general analysis, green for validation, orange for performance/warnings, cyan for style/docs, and purple for architecture/planning.

What tools can I grant to my custom agents?

Agents can be granted specific permissions for tools like Read (viewing files), Grep (searching content), Glob (finding files), Bash (executing commands), and Edit (modifying source code).

Agent Development Toolkit

Name: Agent Development Toolkit
Author: TomazWang

byTomazWang

0•

开发者工具

Streamlines the creation of autonomous, domain-specific agents for Claude Code plugins to automate complex development tasks.

The Agent Development skill provides a comprehensive framework and standardized patterns for building specialized autonomous workers within the Claude Code ecosystem. It guides developers through defining clear agent roles, selecting appropriate permission-based tools, and implementing structured workflows for tasks like code analysis, security auditing, and API validation. By leveraging these established patterns, developers can create reliable background agents that operate independently to deliver high-quality, actionable reports without requiring constant user intervention or interactive control.

主要功能

01Visual identification system using standardized agent color-coding (blue, green, orange, cyan, purple)

020 GitHub stars

03Support for multi-agent coordination through advanced meta-agent patterns

04Standardized templates for Analysis, Validation, Review, and Meta agents

05Comprehensive guidance on optimal tool selection including Read, Grep, Bash, and Task

06Best practices for defining clear agent roles and structured markdown output formats

使用场景

01Building an automated security reviewer to scan for injection vulnerabilities and sensitive data leaks

02Developing background code analyzers to detect anti-patterns and maintain architectural integrity

03Creating a specification validator to ensure OpenAPI or AsyncAPI documentation consistency

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tomazwang/agent-toolkit agent-development

For use in Claude.ai and ChatGPT

Download Skill