Plugin Development

Quick Start

Create a complete agent:

---
description: Expert in X domain, helping with Y and Z
capabilities:
  - "Build X systems"
  - "Optimize performance"
  - "Debug issues"
---

# Agent Name

## Overview
Expert specializing in X domain with 5+ years experience.

## Expert Areas

### Area 1: Core Concepts
Explanation and best practices...

## When to Use
Use this agent when building X or optimizing Y.

## Integration
Works with: agent-2, agent-3, skill-common

Agent Implementation

YAML Frontmatter

---
description: "What agent does. When to use. Max 1024 chars."
capabilities:
  - "Specific capability 1"
  - "Specific capability 2"
  - "Specific capability 3"
  - "Specific capability 4"
---

Content Structure

# Agent Name

## Overview
[1-2 sentences about agent expertise]

## Expert Areas

### Area 1
[Detailed explanation with examples]

### Area 2
[More specific guidance]

### Area 3
[Best practices]

## When to Use
Use this agent when:
- Task 1
- Task 2
- Task 3

## Integration
Works with:
- Agent name (for X)
- Agent name (for Y)
- Skill name (for Z)

---
**Status**: ✅ Production Ready | **Updated**: [Date]

Skill Implementation

SKILL.md Template

---
name: skill-id
description: "What it teaches and when to use (max 1024 chars)"
---

# Skill Name

## Quick Start

[Working code - immediately useful]

```python
# Real example
result = do_something()
print(result)

Core Concepts

Concept 1

[Explanation with code]

Concept 2

[Practical patterns]

Concept 3

[Advanced usage]

Advanced Topics

[Expert-level material]

Real-World Projects

[1-3 practical applications]

Use this skill when:

Learning X
Implementing Y
Solving Z problem


## Command Implementation

### Command Files

```markdown
# /command-name - One-Line Description

## What This Does

[Clear explanation of what command does]

## Usage

/command-name /command-name --option value /command-name --flag1 v1 --flag2 v2


## Options

| Option | Type | Description |
|--------|------|-------------|
| `--option` | string | What it does |
| `--flag` | boolean | Enables X |

## Example

$ /command-name my-plugin Creating plugin... ✅ Done!

Next: /command-2


## Tips

- Tip 1
- Tip 2

## Related Commands

- `/other-command`

Hook Implementation

Hook JSON

{
  "hooks": [
    {
      "id": "unique-id",
      "name": "Hook Display Name",
      "description": "What it does",
      "event": "event-type",
      "condition": "condition-logic",
      "action": "action-handler",
      "enabled": true
    }
  ],
  "notifications": {
    "enabled": true,
    "channels": ["in-app", "console"]
  }
}

Hook Event Types

command-executed - When command runs
agent-invoked - When agent used
skill-loaded - When skill accessed
scheduled - Periodic events

Code Quality Standards

Agent Quality

✅ Clear description (100-200 chars)
✅ 5-10 specific capabilities
✅ 3-5 expert areas
✅ "When to Use" section
✅ Integration points documented
✅ 250-400 lines total

Skill Quality

✅ Name: lowercase-hyphens
✅ Description: actionable, clear
✅ Quick Start: working code
✅ 3+ core concepts
✅ Advanced section
✅ 2+ real projects
✅ 200-300 lines total

Command Quality

✅ Clear description
✅ Usage examples
✅ Options documented
✅ Example output shown
✅ Next steps suggested
✅ 100-150 lines total

Common Implementation Patterns

Knowledge Pattern

Agent → Explains concept
Skill → Provides examples
Command → Enable practice

Workflow Pattern

Command → Starts workflow
Agent → Guides decisions
Hook → Automate steps

Integration Pattern

Agent A → Recommends B
Agent B → Links to skill X
Skill X → Suggests command Y

Testing Your Implementation

Agent Testing

✅ Description under 1024 chars
✅ Capabilities are specific
✅ Content is 250-400 lines
✅ Integration documented
✅ Status included

Skill Testing

✅ Name lowercase-hyphenated
✅ Quick Start runs without error
✅ 3+ concepts explained
✅ Real projects included
✅ Proper formatting

Command Testing

✅ Command executes
✅ Options work as documented
✅ Output matches description
✅ Next steps provided
✅ No errors

Documentation Requirements

For Agents

✅ What agent does
✅ When to use
✅ Capabilities (5-10)
✅ Expert areas (3-5)
✅ Integration points
✅ Status & date

For Skills

✅ Clear description
✅ Quick Start code
✅ Core concepts (3+)
✅ Advanced topics
✅ Real projects (2+)
✅ Usage guidelines

For Commands

✅ What it does
✅ Usage syntax
✅ Options table
✅ Example output
✅ Next steps
✅ Related commands

Version Control Practices

Commit Messages

feat: Add new skill for X
fix: Correct Y in agent
docs: Update Z documentation
refactor: Improve performance
test: Add validation tests

File Changes

New agent?
  → Ensure manifest updated
  → Add to appropriate section
  → Document relationships

New skill?
  → Create folder with SKILL.md
  → Reference in manifest
  → Add to agent capabilities

New command?
  → Create markdown file
  → Add to manifest
  → Document options