Transform your coding workflow with intelligent AI assistance

An advanced Claude Code plugin that brings together slash commands, specialized agents, and intelligent hooks to create a comprehensive AI pair programming experience.

🌟 Features

This plugin integrates three powerful Claude Code features:

📝 Slash Commands

Quick access to AI programming assistance:

• /pair - Start an interactive pair programming session
• /review - Get comprehensive code reviews
• /suggest - Receive intelligent improvement suggestions
• /fix - Automatically detect and fix bugs
• /explain - Get detailed code explanations

🤖 Expert Agents

Specialized AI assistants for different aspects of development:

• @code-reviewer - Security, quality, and best practices expert
• @bug-hunter - Bug detection and fixing specialist
• @architect - Software architecture and design consultant
• @performance-expert - Performance optimization specialist

🔔 Intelligent Hooks

Automatic quality checks throughout your workflow:

• PostToolUse: Suggests code review after significant changes
• PreToolUse: Validates quality before git commits
• SessionEnd: Generates comprehensive session reports

🚀 Installation

Quick Install

# Add the marketplace (if not already added)
/plugin marketplace add https://github.com/Dev-GOM/claude-code-marketplace.git

# Install the plugin
/plugin install ai-pair-programming@dev-gom-plugins

# Restart Claude Code
claude -r

Verify Installation

/plugin

You should see "ai-pair-programming" in the enabled plugins list.

📖 Usage

Starting a Pair Programming Session

/pair

This activates AI pair programming mode where Claude:

• Asks clarifying questions before implementing
• Explains reasoning and design decisions
• Suggests alternatives and trade-offs
• Catches potential mistakes early
• Shares knowledge and best practices

Getting a Code Review

# Review recent changes
/review

# Review a specific file
/review src/components/auth.js

The @code-reviewer agent will analyze:

• Security: Authentication, input validation, vulnerabilities
• Code Quality: Readability, naming, organization
• Performance: Algorithm efficiency, optimization opportunities
• Best Practices: Design patterns, error handling, testing
• Potential Bugs: Edge cases, null checks, logic errors

Requesting Improvement Suggestions

# Get suggestions for recent changes
/suggest

# Get suggestions for a specific file
/suggest src/utils/api.js

Provides:

• Refactoring opportunities
• Design improvements
• Performance optimizations
• Code modernization suggestions
• Maintainability enhancements

Fixing Bugs

# Debug with error message
/fix "TypeError: Cannot read property 'length' of undefined"

# Debug a specific file
/fix src/services/payment.js

The @bug-hunter agent will:

1. Identify the problem from error messages
2. Perform root cause analysis
3. Propose multiple fix approaches
4. Implement the best solution
5. Suggest tests to prevent regression

Getting Code Explanations

# Explain a file
/explain src/hooks/useAuth.ts

# Explain a concept
/explain "how does the authentication flow work"

Provides explanations with:

• Big picture overview
• Component breakdowns
• Execution flow walkthroughs
• Design patterns and best practices
• Concrete examples

Using Expert Agents Directly

You can invoke agents directly in your conversation:

@code-reviewer please review the authentication logic in auth.ts

@bug-hunter there's a memory leak in the WebSocket connection

@architect how should I structure the microservices communication?

@performance-expert the dashboard is loading slowly, can you analyze it?

🔧 How It Works

Intelligent Hook System

The plugin uses hooks to provide proactive assistance:

1. After Writing Code (PostToolUse)

   • Tracks file changes
   • Suggests review after 5 files or 5 minutes
   • Helps maintain consistent quality

2. Before Git Commits (PreToolUse)

   • Checks if code has been reviewed
   • Warns about unreviewed changes
   • Prevents committing problematic code

3. Session End (SessionEnd)

   • Generates summary report
   • Provides statistics
   • Recommends next actions

State Tracking

The plugin maintains state in .state/review-state.json:

{
  "fileChanges": 0,
  "lastReviewTime": 1634567890000
}

This helps track:

• Number of files changed since last review
• Time of last code review
• Session statistics

Generated Reports

At session end, creates .pair-programming-session.md:

# AI Pair Programming Session Report

**Session End:** 2024-10-16 14:30:00

## Session Statistics
- **Files Modified:** 7
- **Last Code Review:** 2024-10-16 14:15:00

## Recommendations
✅ Code is relatively stable. Good time for a commit!

🎯 Workflow Example

Here's a typical workflow using this plugin:

# 1. Start pair programming
/pair
# Claude: "Let's build something great together! What would you like to work on?"

# 2. Implement feature with AI assistance
"Let's add user authentication"
# Claude helps design and implement...

# 3. After 5 files changed, automatic suggestion appears
# 💡 Tip: You've made significant changes. Consider running /review

# 4. Request code review
/review
# @code-reviewer provides detailed analysis

# 5. Fix any issues found
@bug-hunter please fix the JWT validation issue

# 6. Before committing, validation occurs
git commit -m "Add authentication"
# ⚠️ Pre-Commit Check: Run /review to ensure quality

# 7. Final review
/review

# 8. Commit the code
git commit -m "Add authentication"

# 9. End session - automatic report generated
# 🎯 Session Complete! Report saved to .pair-programming-session.md

⚙️ Configuration

Plugin Configuration

The plugin uses hooks/hooks.json for configuration. Key settings:

{
  "configuration": {
    "reviewThresholds": {
      "fileChangeCount": 5,           // Files before suggesting review
      "timeIntervalMinutes": 5,        // Minutes before suggesting review
      "preCommitWarningFiles": 3,      // Files to warn before commit
      "preCommitWarningMinutes": 10    // Minutes to warn before commit
    },
    "outputFiles": {
      "sessionReport": ".pair-programming-session.md",
      "reviewState": ".state/review-state.json"
    },
    "features": {
      "autoReviewSuggestion": true,    // Enable auto-review suggestions
      "preCommitValidation": true,     // Enable pre-commit warnings
      "sessionReporting": true         // Enable session reports
    }
  }
}

Environment Variables:

• ${CLAUDE_PLUGIN_ROOT}: Automatically resolves to plugin root directory
• Used in hook commands to ensure correct script paths

Customizing Agents

Agents are defined as Markdown files with YAML frontmatter in agents/ directory.

Example (agents/code-reviewer.md):

---
name: code-reviewer
description: Your custom description
tools: Read, Grep, Glob
model: sonnet
---

Your custom system prompt goes here...

Agent Properties:

• name: Unique identifier (use lowercase and hyphens)
• description: When to invoke this agent
• tools: Comma-separated list of allowed tools
• model: Model to use (sonnet, opus, haiku, or inherit)

Disabling Specific Hooks

Edit hooks/hooks.json to disable hooks:

{
  "hooks": {
    "PostToolUse": [
      {
        "enabled": false,  // Disable this hook
        ...
      }
    ]
  }
}

Or disable the entire plugin:

/plugin disable ai-pair-programming

📊 Output Files

The plugin generates these files:

• .state/review-state.json - Session state tracking
• .pair-programming-session.md - Session report

Add to .gitignore if desired:

.state/
.pair-programming-session.md

🐛 Troubleshooting

Plugin Not Working

1. Check plugin installation:

   /plugin
   

2. Verify Node.js is available:

   node --version
   

3. Check for errors:

   claude --debug
   

Hooks Not Triggering

1. Ensure hooks are enabled in settings
2. Check hook event names match
3. Verify script permissions (Unix):

   chmod +x plugins/ai-pair-programming/scripts/*.js
   

Agents Not Responding

1. Verify agent Markdown files have valid YAML frontmatter
2. Check agent names match (use @agent-name format)
3. Ensure agents directory is properly configured
4. Confirm agent files use .md extension, not .json

📦 Version

Current Version: 1.1.1

📋 Changelog

v1.1.1 (2025-10-20)

• 🔄 Auto Migration: Plugin version-based configuration migration
• 📦 Smart Updates: Preserves user settings while adding new fields
• 🏷️ Project Scoping: State and output files now use project name to prevent conflicts
• 🎯 SessionStart Hook: Auto-creates configuration file on session start
• ⚡ Performance: SessionStart hook exits immediately if config is up-to-date
• 🌍 Cross-Platform: Enhanced path handling for Windows/macOS/Linux compatibility

v1.0.0

• Initial release

🤝 Contributing

Feel free to customize this plugin:

1. Fork the repository
2. Modify agents, commands, or hooks
3. Test with /plugin validate
4. Share your improvements!

📄 License

Apache License 2.0 - See LICENSE for details

🙏 Credits

Created for the Claude Code community to enhance developer productivity through intelligent AI pair programming.

---

Happy Coding with AI! 🚀

For issues or suggestions, please open an issue on GitHub.