codex-cli
Reference for Codex CLI usage patterns. Consult before calling codex via Bash.
When & Why to Use This Skill
This Claude skill provides a comprehensive reference for the Codex CLI, enabling developers to seamlessly integrate advanced AI capabilities into their Bash workflows. It offers standardized patterns for command execution, session management, and safety configurations, ensuring efficient and secure AI-driven automation directly from the terminal.
Use Cases
- Workflow Automation: Streamline development tasks by incorporating Codex AI prompts into Bash scripts for automated code analysis or generation.
- Secure Environment Management: Utilize sandboxing flags (read-only or workspace-write) to safely execute AI commands while protecting sensitive system files.
- Complex Problem Solving: Configure high-reasoning effort parameters and specific model versions like gpt-5.2 to tackle intricate architectural or debugging challenges.
- Session Persistence: Maintain context across long-running tasks by using session IDs to resume previous AI interactions within the CLI environment.
| name | codex-cli |
|---|---|
| description | Reference for Codex CLI usage patterns. Consult before calling codex via Bash. |
Codex CLI Reference
Quick reference for Codex CLI commands.
Basic Command
codex exec "[prompt]" -m gpt-5.2 --reasoning-effort high --sandbox read-only --ask-for-approval never 2>&1
Common Flags
| Flag | Purpose |
|---|---|
exec |
Non-interactive execution mode |
-m gpt-5.2 |
Model selection |
--reasoning-effort high |
Reasoning effort (minimal, low, medium, high, xhigh) |
--sandbox read-only |
Prevent file modifications |
--sandbox workspace-write |
Allow writes within workspace |
--ask-for-approval never |
Non-interactive mode |
--search |
Enable web search |
-C DIR |
Set working directory |
Session Continuation
# Resume a previous session
codex resume [session-id] "[prompt]" --sandbox read-only --ask-for-approval never 2>&1
Note: Session IDs are returned in Codex output.
Bash Execution Notes
- Use
dangerouslyDisableSandbox: truefor Bash calls - Always append
2>&1to capture all output - Use timeout of 300000ms (5 min) or longer for complex tasks
Safety Requirements
NEVER use these flags:
--dangerously-bypass-approvals-and-sandbox- FORBIDDEN--sandbox danger-full-access- FORBIDDEN
Troubleshooting
Timeout: Complex analysis may take 5-10 minutes. Use longer timeout values.
Authentication: Codex CLI must be authenticated via codex login.
Flag errors: Run codex --help to verify correct flag usage.
More Information
- CLI reference:
codex --help - Official docs: https://github.com/openai/codex