reasoning-modes
Pass
Audited by Gen Agent Trust Hub on Aug 22, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown and YAML configuration files. No scripts, executables, or code snippets are present.
- [SAFE]: The instructions focus on conversational styling and analytical frameworks (e.g., 'brainstorming', 'adversarial review'). There are no commands that interact with the file system, network, or environment variables.
- [PROMPT_INJECTION]: The instructions include safety-conscious language such as 'Change the interaction pattern without changing truth, safety, permissions' and 'Do not broaden authorization'. It specifically warns against ignoring safety protocols.
Audit Metadata