codex-skill

Run OpenAI Codex CLI commands for automated code refactoring by executing `codex exec` with specific flags for model selection and reasoning effort. The Skill manages sandbox environments and integrates user prompts to define interactive code analysis and editing tasks directly within your workflow.

Yes, you can resume a previous Codex CLI session to continue code editing tasks by using the `codex resume` command. This allows you to pick up interactive task definitions and continue modifying code with your previously selected model and reasoning effort settings.

The sandbox modes available for Codex code generation and analysis are read-only, workspace-write, and full-access. These sandbox environments manage the level of file modification permitted during automated editing, ensuring safe execution of refactoring tasks according to your access requirements.

Configure reasoning effort and model selection in Codex CLI by specifying the appropriate flags when running `codex exec`. You can set high reasoning effort for complex refactoring tasks or choose different models to balance processing speed and code analysis depth.

Yes, Codex CLI supports automated code analysis in a read-only sandbox. You can execute `codex exec` with a read-only sandbox configuration to safely analyze and review code without risking unintended modifications to your workspace files.

Limitations when using Codex CLI for automated editing tasks include dependency on available sandbox environments and reasoning effort configurations. Full-access sandbox modes carry higher risk of unintended file modifications, requiring careful prompt definition for complex code refactoring operations.