behavioral-modes
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structural templates for different AI interaction styles. It focuses on task prioritization and communication preferences rather than expanding the agent's permissions or capabilities.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No exfiltration patterns were detected. In 'SHIP' mode, the skill explicitly instructs the agent to check for exposed secrets and environment configuration issues, which is a defensive security measure.
- [PROMPT_INJECTION]: Instructions within the 'IMPLEMENT' mode to 'minimize questions' and avoid 'verbose explanations' are standard performance preferences for code generation and do not attempt to bypass core safety guidelines or agent constraints.
- [INDIRECT_PROMPT_INJECTION]: While the skill defines modes for processing external data (e.g., 'REVIEW', 'DEBUG', 'SHIP'), it implements structured output templates that help the agent maintain a consistent persona. Additionally, the 'SHIP' and 'REVIEW' modes explicitly mandate security-focused checkpoints which mitigates the risk of ignoring malicious content embedded in user data.
Audit Metadata