metahuman-identity-pipeline
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a legitimate workflow for MetaHuman creation using Unreal Engine. The Python scripts provided are focused on technical validation and automation of facial markers and rigging.
- [REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as piping network downloads to a shell or using dangerous eval/exec functions with external data, were detected.
- [DATA_EXPOSURE_AND_EXFILTRATION]: There are no hardcoded credentials, sensitive file path accesses, or unauthorized network operations. The
api_keyargument in the pipeline class is a standard placeholder and does not contain sensitive data. - [OBFUSCATION]: The skill does not use Base64 encoding, zero-width characters, homoglyphs, or any other techniques to hide malicious instructions or URLs.
- [INDIRECT_PROMPT_INJECTION]: While the skill ingests external data in the form of photos, the processing is performed using algorithmic computer vision (OpenCV) and the Unreal Engine API, which does not interpret the content as natural language instructions, effectively mitigating the risk of indirect injection.
Audit Metadata