What is Codex autonomous execution mode?

It is a non-interactive automation mode that allows the agent to execute tasks from start to finish without seeking approval for every individual action, focusing on efficiency and results.

Can I use the Codex skill for non-Git projects?

Yes, by using the --skip-git-repo-check flag, Codex can analyze and modify directories that are not initialized as Git repositories.

What happens if a task is interrupted?

The skill includes a 'resume' command that allows you to pick up the last session and continue with the workflow exactly where it stopped.

How do sandbox modes work in Codex?

Codex offers three levels: Read-Only for analysis, Workspace-Write for standard programming within a project, and Danger-Full-Access for tasks requiring network or system-level permissions.

Which models are supported by the Codex skill?

The skill supports the GPT-5 and GPT-5.1 model families, including specialized versions like gpt-5.1-codex for optimized programming performance.

Codex Autonomous Execution

Name: Codex Autonomous Execution
Author: Microck

byMicrock

•

생산성 및 워크플로

Automates complex coding tasks and feature implementations using Codex CLI for hands-off, non-interactive execution.

소개

The Codex Skill empowers Claude to delegate execution tasks to the Codex CLI, enabling a truly autonomous, non-interactive workflow. By leveraging advanced models like GPT-5 and GPT-5.1, it can implement features, refactor codebases, and fix bugs across various sandbox environments without requiring constant user approval for every sub-step. This skill is ideal for developers who have established a technical plan and want an AI agent to handle the heavy lifting of implementation while providing structured progress updates and comprehensive final summaries.

주요 기능

89 GitHub stars
Granular sandbox security modes including Read-Only and Workspace-Write
Structured JSON output for detailed metrics and change tracking
Session resumption capabilities to continue long-running workflows
Non-interactive autonomous execution for hands-off task completion
Support for GPT-5 and GPT-5.1 model selection for high-reasoning tasks

사용 사례

Performing large-scale codebase refactoring without manual intervention
Running autonomous security audits and vulnerability scans with reporting
Implementing complex features based on predefined design documents

소개

주요 기능

89 GitHub stars
Granular sandbox security modes including Read-Only and Workspace-Write
Structured JSON output for detailed metrics and change tracking
Session resumption capabilities to continue long-running workflows
Non-interactive autonomous execution for hands-off task completion
Support for GPT-5 and GPT-5.1 model selection for high-reasoning tasks

사용 사례

Performing large-scale codebase refactoring without manual intervention
Running autonomous security audits and vulnerability scans with reporting
Implementing complex features based on predefined design documents