Codex Delegator - Claude Code Agent Skill

Enables Claude Code to automatically delegate complex tasks to OpenAI Codex CLI for AI-to-AI collaboration

What is This?

Codex Delegator is a Claude Code Agent Skill - a specialized module that extends Claude Code with the ability to automatically delegate complex, logic-intensive programming tasks to OpenAI's Codex CLI.

Think of it as giving Claude Code a specialized co-worker (Codex) for handling particularly challenging problems.

What It Does

When you're working with Claude Code and encounter:

• 🔧 Complex backend logic implementations
• 🧮 Intricate algorithm optimizations
• 🐛 Persistent bugs that resist multiple fix attempts
• 🔄 Race conditions and concurrency issues
• 💡 Problems needing a fresh perspective

Claude Code can automatically invoke Codex to provide alternative solutions, leveraging the strengths of both AI systems.

How It Works

You: "This payment processing bug keeps happening under high load,
     I've tried 3 different fixes"

  ↓

Claude Code: [Analyzes problem, recognizes delegation criteria]
"I'm going to use Codex to help debug this race condition..."

  ↓

Codex: [Analyzes with fresh perspective, generates solution]

  ↓

Claude Code: [Validates solution, runs tests, integrates]
"Here's the fix. The issue was a race condition in Redis cache updates..."

Installation

Prerequisites

1. Claude Code - Install from claude.com/claude-code
2. OpenAI Codex CLI - Install globally:

   npm i -g @openai/codex

3. Authentication - ChatGPT Plus/Pro/Business/Enterprise account OR OpenAI API key:

   codex auth

Install the Skill

Option 1: Download and Extract

1. Download codex-delegator.zip
2. Extract to ~/.claude/skills/codex-delegator
3. Restart Claude Code

Option 2: Git Clone (if developing)

cd ~/.claude/skills
git clone https://github.com/your-org/codex-delegator.git

Usage Examples

Example 1: Persistent Bug

You: "This race condition in my payment system has been
     failing 5% of transactions. I've tried adding
     transactions, increasing Redis TTL, and retry logic."

Claude Code: "I'm delegating this to Codex for a fresh
             analysis of the race condition..."

[Codex analyzes and fixes]

Claude Code: "Found the issue - your webhook handler and
             order service were competing for cache locks.
             Here's the fix using distributed locking..."

Example 2: Algorithm Optimization

You: "My recommendation engine is too slow - 30 seconds
     for 10k items. Need to handle 100k+ items."

Claude Code: "This is a complex optimization problem.
             Using Codex to explore algorithmic improvements..."

[Codex optimizes]

Claude Code: "Reduced complexity from O(n²) to O(n log n)
             using spatial indexing. Now processes 100k
             items in under 1 second."

Configuration

The skill uses intelligent decision-making to determine when delegation is beneficial. It automatically delegates when:

• ✅ Problem attempted 2+ times without resolution
• ✅ High logic complexity (nested conditions, complex state)
• ✅ Backend/algorithm heavy tasks
• ✅ Need for different problem-solving approach

What's Inside?

codex-delegator/
├── SKILL.md                      # Main skill documentation
├── references/
│   └── execution_strategies.md   # Detailed Codex usage patterns
└── README.md                     # This file

Requirements

• Claude Code: Latest version recommended
• Codex CLI: v0.58.0+
• Node.js: For Codex CLI installation
• Git repository: Codex works best in version-controlled projects
• OpenAI Account: ChatGPT Plus/Pro/Business/Enterprise OR API key

Limitations

This skill is designed for complex problems. It's not used for:

• Simple CRUD operations
• Basic UI components
• Straightforward bug fixes (first attempt)
• Simple configuration changes
• General coding questions

About Agent Skills

Agent skills are modular packages that extend Claude Code with specialized capabilities. They work by:

1. Providing specialized knowledge - Domain-specific workflows and best practices
2. Tool integration - Connecting Claude Code with external tools (like Codex)
3. Automated decision-making - Knowing when to apply specialized capabilities

Learn more about creating agent skills: Claude Code Documentation

Contributing

Contributions welcome! This skill can be improved with:

• Additional delegation strategies
• More example scenarios
• Better decision-making criteria
• Performance optimizations

See CONTRIBUTING.md for guidelines.

License

MIT License - see LICENSE file for details.

Acknowledgments

• Built for Claude Code by Anthropic
• Uses OpenAI Codex CLI
• Inspired by the idea of AI collaboration and specialization

Support

• 📖 Documentation
• 🐛 Report Issues
• 💬 Discussions

---

Made with ❤️ for the Claude Code community