calibrated-assertiveness
Pass
Audited by Gen Agent Trust Hub on Jul 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill consists entirely of instructional text and behavioral guidelines. It does not include any executable code, shell commands, or network operations.
- [PROMPT_INJECTION]: The instructions encourage the agent to be more direct and avoid 'defensive downgrading' or 'alignment tax.' While this modifies the agent's default response style, the skill explicitly states that genuine risks should still be factored into decisions, ensuring that the model remains aware of actual safety concerns rather than generic caveats.
- [EXTERNAL_DOWNLOADS]: No external resources, packages, or remote scripts are requested or downloaded.
- [DATA_EXFILTRATION]: No mechanisms for accessing sensitive data or exfiltrating information were detected. References to vendor-specific feedback tools (e.g., 'report-biaoo-skill-feedback') and repositories ('Biaoo/skills') are used for legitimate skill maintenance.
Audit Metadata