inspect-fusion-design
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill processes external data from Fusion design files, such as component names, comments, and property text. The instructions proactively address the risk of indirect prompt injection by stating that these fields should be treated as untrusted data rather than instructions. This approach helps ensure that the agent remains focused on the engineering analysis rather than being influenced by text embedded within the models.
- Data Handling and Tool Scope: The skill utilizes specialized tools for geometry measurement and design checking. The workflow is restricted to read-only inspection and model understanding, without involving operations that would allow for unauthorized data modification or exfiltration.
Audit Metadata