wildworld-dataset
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user to clone the official project repository from GitHub (
https://github.com/ShandaAI/WildWorld.git). This is a standard procedure for accessing research datasets and code. - [COMMAND_EXECUTION]: Includes standard commands for repository setup (
git clone,cd,pip install). These are routine development tasks and do not involve unauthorized privilege escalation or hidden execution. - [DATA_EXFILTRATION]: The provided Python code reads local JSON and image files from the dataset directory. This is the intended primary purpose of the skill (dataset loading) and does not involve sending sensitive user data to external servers.
- [PROMPT_INJECTION]: No prompt injection or behavior override patterns were detected in the instructions or metadata.
Audit Metadata