agent-development

Installation
Summary

Comprehensive guide for building autonomous agents in Claude Code plugins with structured frontmatter, system prompts, and triggering conditions.

  • Agents are autonomous subprocesses defined in markdown files with YAML frontmatter; use them for multi-step independent work, not user-initiated commands
  • Required frontmatter fields: name (3–50 lowercase-hyphenated characters), description (with 2–4 concrete triggering examples), model (inherit/sonnet/opus/haiku), and color (blue/cyan/green/yellow/magenta/red)
  • System prompts should follow a structured template with clear responsibilities, step-by-step analysis process, quality standards, output format, and edge case handling; keep under 10,000 characters
  • Optional tools field restricts agent access to specific capabilities (e.g., Read, Write, Grep, Bash) following the principle of least privilege
  • Includes validation rules for identifiers and descriptions, testing strategies for triggering and system prompt behavior, and utility scripts for validation and testing
SKILL.md

Agent Development for Claude Code Plugins

Overview

Agents are autonomous subprocesses that handle complex, multi-step tasks independently. Understanding agent structure, triggering conditions, and system prompt design enables creating powerful autonomous capabilities.

Key concepts:

  • Agents are FOR autonomous work, commands are FOR user-initiated actions
  • Markdown file format with YAML frontmatter
  • Triggering via description field with examples
  • System prompt defines agent behavior
  • Model and color customization

Agent File Structure

Complete Format

---
Related skills

More from anthropics/claude-plugins-official

Installs
2.5K
GitHub Stars
19.2K
First Seen
Feb 5, 2026