Tag
The author open-sourced a Codex skill called 'prompt-to-loop-engineer' that converts vague prompts into structured, intent-preserving execution loops with anti-drift checks, aiming to make agents more reliable for real-world tasks.