Codex Skill

Defaults

Do NOT ask the user for model or reasoning effort unless explicitly requested.

Running a Task

Sandbox Mode Selection

Command

codex exec --skip-git-repo-check \
  -m gpt-5.3-codex \
  --config model_reasoning_effort="high" \
  --sandbox read-only \
  "your prompt here" 2>/dev/null

Always:

• Use --skip-git-repo-check
• Append 2>/dev/null to suppress thinking tokens (show stderr only when debugging)
• Confirm with user before using --full-auto or --sandbox danger-full-access

Resuming a Session

echo "your prompt here" | codex exec --skip-git-repo-check resume --last 2>/dev/null

All flags go between exec and resume. Omit configuration flags when resuming unless the user requests them.

After Completion

• Inform the user: "You can resume this Codex session by saying 'codex resume'."
• Use AskUserQuestion to confirm next steps or whether to resume.
• When output includes warnings or partial results, summarize and ask how to adjust.

Critical Evaluation

Treat Codex as a colleague, not an authority.

• Trust your knowledge when confident. Push back on incorrect claims directly.
• Research disagreements using WebSearch or docs before accepting Codex's claims.
• Remember knowledge cutoffs — Codex may not know about recent releases or API changes.

When Codex is Wrong

1. State your disagreement clearly to the user
2. Provide evidence (your knowledge, web search, docs)
3. Optionally resume to discuss:

   echo "This is Claude following up. I disagree with [X] because [evidence]. What's your take?" | codex exec --skip-git-repo-check resume --last 2>/dev/null
   

4. Frame disagreements as discussions — either AI could be wrong
5. Let the user decide when there's genuine ambiguity

Error Handling

• Report failures whenever codex --version or codex exec exits non-zero.
• When output includes warnings or partial results, summarize and ask how to adjust.