paper-section-author
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses the
disable-model-invocation: trueconfiguration and explicit internal rules to prevent the agent from executing commands or calling external tools. - [SAFE]: It includes specific instructions for escaping LaTeX control characters in prose, which mitigates risks associated with malformed output or unintended command execution in downstream compilers.
- [SAFE]: No indicators of prompt injection, data exfiltration, or persistence mechanisms were detected in the analyzed file.
Audit Metadata