Agent skill

codex-agent-collaboration

Delegate coding tasks to Codex AI for implementation, analysis, and alternative solutions. Use when you need a second AI perspective, want to explore different approaches, or need specialized Codex capabilities for complex coding tasks.

Stars 25
Forks 2

Install this agent skill to your Project

npx add-skill https://github.com/ForteScarlet/codex-kkp/tree/main/claude-code-skills-template/codex-agent-collaboration

SKILL.md

Codex CLI Skill

This skill enables Claude Code to execute tasks using OpenAI's Codex AI agent.

Overview

The codex-kkp-cli is a Codex Agent CLI tool, allowing you to:

  • Execute coding tasks and get implementations
  • Perform code analysis and reviews
  • Get alternative solutions and suggestions
  • Collaborate with Codex for cross-checking implementations

Usage

Basic Syntax

bash
# Direct call with platform-specific executable
executables/codex-kkp-cli-{platform} --cd=/absolute/path/to/project [options] "<task_description>"

Where {platform} is one of:

  • macosx64 - macOS Intel (x86_64)
  • macosarm64 - macOS Apple Silicon (ARM64)
  • linuxx64 - Linux x86_64
  • linuxarm64 - Linux ARM64
  • mingwx64 - Windows x86_64

Platform Auto-Detection Helper: A platform detection script is provided to help identify your current platform:

On Windows, Just use mingwx64 platform directly, no need to use script detection.

bash
# Unix/Linux/macOS
codex-kkp-cli-platform
# Outputs: macosx64, macosarm64, linuxx64, or linuxarm64

communication

This is AI-to-AI communication between You and Codex. PRIORITIZE ACCURACY AND PRECISION over human readability. Use structured data, exact technical terms, full paths, and precise details. NO conversational formatting needed.

Required Parameters

Parameter Description
Task The task description (positional argument, must be quoted)
--cd=<dir> Working directory (ABSOLUTE PATH REQUIRED)

Optional Parameters

Parameter Description
--session=<id> Session ID (STRONGLY RECOMMENDED for follow-up chats to maintain context)
--sandbox=<mode> Sandbox mode. Default is read-only. See sandbox-modes.md
--full-auto Allow Codex to edit files automatically
--image=<path> Include an image file (ABSOLUTE PATH, can repeat)
--skip-git-repo-check[=BOOL] Skip Git repository check. Default is true. Use =false to enable Git check

For output options (--full, --output-last-message, --output-schema), see outputs.md.

NOTE that parameters and values are connected by an EQUAL SIGN =, not a space.

Response Format

Returns JSON with "type": "SUCCESS" or "type": "ERROR".

JSON
{
  "type": "SUCCESS",
  "session": "xxxxxxx",
  "content": {
    "agentMessages": "I've analyzed the code and found...",
    "fileChanges": [...],   // Optional
    "nonFatalErrors": [...] // Optional
  }
}
  • fileChanges and nonFatalErrors is nullable.
  • Error responses do NOT include a session field.

Quick Example

New Session:

bash
executables/codex-kkp-cli-{platform} --cd=/path/to/project "Explain the main function in Main.kt"

Continue Previous Session:

bash
executables/codex-kkp-cli-{platform} --cd=/path/to/project --session=xxxxxxx "Explain the main function in Main.kt"

More examples: examples.md

Expand your agent's capabilities with these related and highly-rated skills.

davila7/claude-code-templates

verl-rl-training

Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.

23,776 2,298
Explore
davila7/claude-code-templates

openrlhf-training

High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.

23,776 2,298
Explore
davila7/claude-code-templates

gguf-quantization

GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.

23,776 2,298
Explore
davila7/claude-code-templates

Claude Code Guide

Master guide for using Claude Code effectively. Includes configuration templates, prompting strategies "Thinking" keywords, debugging techniques, and best practices for interacting with the agent.

23,776 2,298
Explore
davila7/claude-code-templates

qdrant-vector-search

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

23,776 2,298
Explore
davila7/claude-code-templates

behavioral-modes

AI operational modes (brainstorm, implement, debug, review, teach, ship, orchestrate). Use to adapt behavior based on task type.

23,776 2,298
Explore

Didn't find tool you were looking for?

Be as detailed as possible for better results