Contributing Guide

Contributing Guide

Guide for creating new plugins and contributing to Arkhe Claude Plugins.

Creating a New Plugin

1. Create Plugin Directory Structure

mkdir -p plugins/new-plugin/.claude-plugin
mkdir -p plugins/new-plugin/agents
mkdir -p plugins/new-plugin/commands
mkdir -p plugins/new-plugin/skills

2. Create Plugin Manifest

Create `plugins/new-plugin/.claude-plugin/plugin.json`:

{
  "name": "new-plugin",
  "description": "Brief description of your plugin's purpose",
  "version": "1.0.0",
  "author": {
    "name": "Your Name"
  }
}

3. Add to Marketplace

Edit `.claude-plugin/marketplace.json` and add your plugin:

{
  "name": "new-plugin",
  "source": "./plugins/new-plugin",
  "description": "Brief description of your plugin's purpose"
}

4. Create README

Create `plugins/new-plugin/README.md` with:

• Overview
• Components (agents, commands, skills)
• Installation instructions
• Usage examples
• Version information

---

Component Guidelines

Creating Agents

Create agent files in `agents/` directory with YAML frontmatter:

---
name: agent-name
description: When and why to use this agent (include trigger phrases like "Use PROACTIVELY")
tools: Read, Write, Grep, Glob, Bash  # Optional - omit to inherit all
model: sonnet  # Optional - sonnet/opus/haiku or inherit
---

System prompt defining the agent's role, capabilities, and approach.

**Naming:** Use lowercase with hyphens (e.g., `code-explorer`, `ai-engineer`)

**Description Guidelines:**

• Keep under 1,024 characters
• Include specific trigger phrases
• Explain when to use the agent

Creating Commands

Create command files in `commands/` directory:

---
description: Brief description of what this command does
argument-hint: [optional: argument description]
---

# Command Name

Full prompt that Claude Code will execute when the command is invoked.
Include specific instructions, context, and expected behavior.

**Naming:** Use lowercase with hyphens (e.g., `code-explain`, `create-pr`)

Creating Skills

Create skills in `skills/skill-name/` directory with progressive disclosure:

skills/skill-name/
├── SKILL.md              # Main instructions (<150 lines)
├── WORKFLOW.md           # Detailed steps (optional)
├── EXAMPLES.md           # Usage examples (recommended)
├── TROUBLESHOOTING.md    # Error handling (recommended)
└── scripts/              # Python scripts (optional)

**SKILL.md Template:**

---
name: Skill Name
description: What it does. Use when [trigger scenarios].
---

# Quick Start
Essential instructions only (<150 lines)

## Output Structure
What the skill produces

## Common Issues
Quick fixes with references to TROUBLESHOOTING.md

## Examples
Brief examples with reference to EXAMPLES.md

**Key Guidelines:**

• SKILL.md: <5,000 tokens (target: <1,000 tokens, ~150 lines)
• Supporting docs: Unlimited size
• Include specific trigger keywords in descripti