threejs-glb-audit
Warn
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill instructions explicitly direct the execution of the command
npx -y paprika3d serve. The-yflag innpxbypasses user confirmation to download and execute the package directly from the npm registry, which is a mechanism that can be leveraged for remote code execution. - [EXTERNAL_DOWNLOADS]: The skill relies on an external tool and package named
paprika3dand directs users to the domainpaprika3d.com. This package and domain are not associated with the skill author (keeno-corp) or any established trusted organizations. - [INDIRECT_PROMPT_INJECTION]: The skill audits 3D model files, creating an attack surface for indirect prompt injection through malicious metadata.
- Ingestion points: The skill reads data from
.glb,.gltf, and.usdfiles (SKILL.md). - Boundary markers: The instructions do not define boundary markers or instruct the agent to ignore instructions embedded within the model metadata.
- Capability inventory: The skill employs the
model_audittool and encourages the execution of shell commands vianpx(SKILL.md). - Sanitization: There is no evidence of metadata sanitization or filtering before the data is processed by the agent.
Audit Metadata