Codex 0.146.1 Applies Safer Automatic-Review Defaults Specifically for Cyber-Capable Models
Codex Changelog·medium signal
The August 5 Codex CLI 0.146.1 release was a single fix: safer automatic-review defaults for cyber-capable models, plus clearer explanation of permission changes in the terminal. This is a model-class-conditional permission default — the harness now varies its own guardrails based on which model is driving it, rather than applying one policy across the board. Expect this to become standard as labs ship models with explicit cyber capability ratings.