openscad-iterative-modeling
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user inputs, such as design specifications and images, to generate executable OpenSCAD code which is then processed by rendering and export tools. \n
- Ingestion points: User requests, object images, and dimensions provided during the "Normalize the request" and "Visually critique" phases in
SKILL.md. \n - Boundary markers: Absent; the instructions do not define clear delimiters or "ignore embedded instruction" warnings for the data being modeled. \n
- Capability inventory: The skill leverages
render_scad_pngfor visualization andexport_scad_stlfor writing geometry files to the filesystem. \n - Sanitization: Absent; the skill relies on the agent's interpretation of user intent to generate code without explicit validation or escaping of the input source material.
Audit Metadata