codex-zdr
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- COMMAND_EXECUTION (HIGH): The skill requires the use of the
Bashtool and provides examples of executing a script namedcodex-yolo.shwith high-privilege flags like--full-auto. The 'yolo' designation often implies bypassing safety checks. - EXTERNAL_DOWNLOADS (HIGH): The skill references a local script file
scripts/multi-model/codex-yolo.shthat is not provided in the skill package. This represents an unverified dependency with unknown behavior. - DATA_EXFILTRATION (MEDIUM): The 'Library-First Protocol' directs the agent to scan and 'extract' code from
D:\Projects\*. This allows the agent to read potentially sensitive files across the entire drive. - PROMPT_INJECTION (HIGH): As a Category 8 risk, the skill processes external code files from
D:\Projects\*while possessingBashexecution capabilities. It lacks boundary markers or sanitization, making it highly vulnerable to instructions embedded in the files it reads.
Recommendations
- AI detected serious security threats
Audit Metadata