Execute Codex CLI tasks to analyze, refactor, or edit code.
npx skills add https://github.com/dayoumin/Skills --skill codex-dayoumin
Please help me install this Agent Skill. Skill: codex Source: https://github.com/dayoumin/Skills/tree/main/skills/codex Command: npx skills add https://github.com/dayoumin/Skills --skill codex-dayoumin
This Skill provides direct access to OpenAI's Codex CLI, enabling advanced code analysis, refactoring, and automated editing. It streamlines complex coding tasks by leveraging a powerful AI model, reducing manual effort and potential errors.
read-only, workspace-write, or danger-full-access to control AI permissions.read-only mode, then, if approved, "Apply the suggested fixes" using workspace-write mode.Ask Claude to "Run Codex to analyze the 'main.py' file for security vulnerabilities using gpt-5-codex and high reasoning effort."
High-intent search queries and answers about installing and using this skill.
Code analysis with Codex CLI involves running `codex exec` with your file and specifying `read-only` mode to prevent unwanted changes. You select the model (e.g., `gpt-5-codex`), set reasoning effort (`high` for thorough review), and Codex scans for bugs, performance problems, or security vulnerabilities without modifying your code.
Yes. After analyzing code in read-only mode, use `codex exec` with `workspace-write` mode to apply refactoring suggestions directly. Codex implements the changes locally while preserving your project structure, making it practical for large-scale refactoring without manual editing.
Sandbox modes control Codex permissions: `read-only` prevents any modifications and is safe for analysis; `workspace-write` allows edits to your files and is suitable for refactoring; `danger-full-access` grants unrestricted permissions for complex automated tasks but requires careful oversight.
Use the `codex resume` command to continue from where a previous `codex exec` session stopped. This preserves context and reasoning from the interrupted task, avoiding redundant analysis and reducing compute overhead.
Codex supports multiple models including `gpt-5-codex` and reasoning effort levels like `high` for intensive analysis. The metadata specifies model selection and reasoning effort as functional requirements you configure per task to balance thoroughness and execution time.
Yes. Codex operates on local projects, so you need your code files accessible in your workspace. The Skill applies read-only or write-enabled sandbox modes based on your chosen permissions, but the project structure and files must already be present.
fmindAutomate MLOps project bootstrap with uv, git, and VS Code configuration.
u9401066Update project documentation before Git commits with semantic versioning.
ricableFacilitate AI-assisted pair programming with dynamic role management and automated verification.
sandalsoftAutomate code analysis, refactoring, and editing with Codex CLI and GPT-5.2.