Does this skill provide guidance on prompting frontier models?

Yes, it includes built-in reference notebooks for GPT-5.2 and GPT-5.1-Codex-Max that provide best practices for verbosity control and tool usage.

What models does the Codex skill support?

It supports frontier models including gpt-5.2-codex, gpt-5.1-codex-max, and the standard gpt-5.2 for a variety of reasoning and coding tasks.

How can I control the depth of the AI's reasoning?

When starting a task, you can specify the reasoning effort as medium, high, or xhigh to balance between execution speed and deep problem-solving.

How do I resume a session in Codex?

You can use the 'codex resume' command or ask Claude to 'continue with codex' to pick up from the last session using the stored state.

Is it safe to run Codex on my local codebase?

Yes, the skill includes mandatory sandbox management, allowing you to restrict Codex to read-only mode or grant specific workspace-write permissions as needed.

Codex CLI Integration

Name: Codex CLI Integration
Author: jongwony

byjongwony

Developer Tools

Delegates complex coding and reasoning tasks to the OpenAI Codex CLI with advanced model selection and session management.

About

The Codex skill enables Claude to interface directly with the OpenAI Codex CLI, allowing for the delegation of high-level agentic coding tasks and deep reasoning sessions to frontier models. It provides a robust framework for managing model selection (including GPT-5.2 and Codex-Max), reasoning effort levels, and security through granular sandbox configurations. By leveraging built-in reference guides and session resumption capabilities, this skill ensures Claude can perform end-to-end development tasks with high precision and developer control.

Key Features

Secure sandbox management with read-only, workspace-write, and full-access modes
Integrated prompting guides for optimized behavior and migration strategies
Configurable reasoning effort levels ranging from medium to extra-high
Session resumption features to continue or iterate on previous Codex tasks
0 GitHub stars
Multi-model support for GPT-5.2, GPT-5.1-Codex-Max, and specialized Codex variants

Use Cases

Delegating agentic coding tasks that require specific sandbox permissions and file system interaction
Executing deep architectural analysis or complex refactoring using high-reasoning frontier models
Managing long-running development sessions that require context persistence and model-specific optimization

About

Key Features

Secure sandbox management with read-only, workspace-write, and full-access modes
Integrated prompting guides for optimized behavior and migration strategies
Configurable reasoning effort levels ranging from medium to extra-high
Session resumption features to continue or iterate on previous Codex tasks
0 GitHub stars
Multi-model support for GPT-5.2, GPT-5.1-Codex-Max, and specialized Codex variants

Use Cases

Delegating agentic coding tasks that require specific sandbox permissions and file system interaction
Executing deep architectural analysis or complex refactoring using high-reasoning frontier models
Managing long-running development sessions that require context persistence and model-specific optimization