01Automated code refactoring and editing via codex exec
02Granular sandbox security modes including read-only and workspace-write
03Session persistence and resumption with context-aware resume commands
04Configurable reasoning effort levels (low, medium, high)
05Support for specialized gpt-5.1-codex model variations
061 GitHub stars