Using Openai Codex CLI

Codex Skill Guide

Running a Task

1. Ask the user (via AskUserQuestion) which model to run (gpt-5 or gpt-5-codex) AND which reasoning effort to use (low, medium, or high) in a single prompt with two questions.
2. Select the sandbox mode required for the task; default to --sandbox read-only unless edits or network access are necessary.
3. Assemble the command with the appropriate options:
   • -m, --model <MODEL>
   • --config model_reasoning_effort="<low|medium|high>"
   • --sandbox <read-only|workspace-write|danger-full-access>
   • --full-auto
   • -C, --cd <DIR>
   • --skip-git-repo-check
4. Always use --skip-git-repo-check.
5. When continuing a previous session, use codex exec --skip-git-repo-check resume --last via stdin. When resuming don't use any configuration flags unless explicitly requested by the user e.g. if he species the model or the reasoning effort when requesting to resume a session. Resume syntax: echo "your prompt here" | codex exec --skip-git-repo-check resume --last 2>/dev/null. All flags have to be inserted between exec and resume.
6. IMPORTANT: By default, append 2>/dev/null to all codex exec commands to suppress thinking tokens (stderr). Only show stderr if the user explicitly requests to see thinking tokens or if debugging is needed.
7. Run the command, capture stdout/stderr (filtered as appropriate), and summarize the outcome for the user.
8. After Codex completes, inform the user: "You can resume this Codex session at any time by saying 'codex resume' or asking me to continue with additional analysis or changes."

Quick Reference

Code Review Workflow

0. IMPORTANT/DO NOT SKIP: Always run the command codex-code-review --help and read the help message before using the script.
1. Determine what to review:
   • Staged/unstaged changes: Use codex-code-review "<request>" [files...]
   • Committed changes: Use codex-code-review --diff "<range>" "<request>" [files...]
     • Examples: --diff "main...HEAD" (PR changes), --diff "HEAD~3..HEAD" (last 3 commits)
2. The script automatically:
   • Uses --sandbox read-only (reviews don't modify files)
   • Includes staged and unstaged git changes (or committed with --diff)
   • Uses gpt-5-codex model with medium reasoning effort (hardcoded)
   • Suppresses thinking tokens for clean output
   • Uses --skip-git-repo-check flag
   • Runs in foreground (do NOT run in background - output won't be captured)
3. After completion, inform user: "You can resume this Codex session for follow-up analysis using: echo 'your question' | codex exec --skip-git-repo-check resume --last 2>/dev/null"
4. Optional: If user wants to resume, pipe their question directly to the resume command

Note:

• The script reviews ALL changes (both staged and unstaged) by default
• For committed changes (e.g., PR reviews), use --diff option: codex-code-review --diff "main...HEAD" "<request>"
• Users can stage specific files first if they want to review only certain changes
• CRITICAL: Always run in foreground - background execution will not capture output properly

Following Up

• After every codex command, immediately use AskUserQuestion to confirm next steps, collect clarifications, or decide whether to resume with codex exec resume --last.
• When resuming, pipe the new prompt via stdin: echo "new prompt" | codex exec resume --last 2>/dev/null. The resumed session automatically uses the same model, reasoning effort, and sandbox mode from the original session.
• Restate the chosen model, reasoning effort, and sandbox mode when proposing follow-up actions.

Error Handling

• Stop and report failures whenever codex --version or a codex exec command exits non-zero; request direction before retrying.
• Before you use high-impact flags (--full-auto, --sandbox danger-full-access, --skip-git-repo-check) ask the user for permission using AskUserQuestion unless it was already given.
• When output includes warnings or partial results, summarize them and ask how to adjust using AskUserQuestion.

Additional References

• Codex CLI Help Menu (headless execution) - Complete help output showing all available options and flags for codex exec non-interactive execution
• Codex CLI Headless Execution Mode - Overview of non-interactive execution for automation workflows, including CLI arguments, sandbox policies, and session resumption
• Codex CLI Passing Inputs and Prompts - Guide to attaching files/images, using custom prompt templates, and passing prompts via arguments or stdin
• Codex CLI Slash Commands and Features - Summary table of interactive slash commands (e.g., /model, /approvals, /review) available during Codex sessions
• Codex CLI Subcommands and Entry Points - CLI architecture documentation covering subcommand dispatch, authentication, execution modes, and utility commands