huashu-nuwa
Fail
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: CRITICAL
Full Analysis
- [SAFE]: The skill is a complex instructional framework for creating other skills. It does not contain any executable scripts, binary files, or commands that perform network or file system operations outside of its documented purpose.
- [EXTERNAL_DOWNLOADS]: The skill references several external URLs for research purposes and provides scripts (e.g.,
fetch_youtube_subtitles.sh) that useyt-dlpto download subtitles. These operations are transparent, targeted at well-known services (YouTube), and consistent with the skill's purpose of gathering research material. - [PROMPT_INJECTION]: While the skill contains detailed instructions on how the agent should think and speak when emulating a personality, it does not attempt to bypass safety guidelines or override systemic constraints. Phrases like "Ignore previous instructions" are not present.
- [DATA_EXFILTRATION]: There are no patterns suggesting the collection or transmission of sensitive data. The skill focuses on public domain information about public figures.
- [COMMAND_EXECUTION]: The skill includes shell scripts and Python scripts for data processing (cleaning subtitles, analyzing titles). These are utility scripts for the user/agent to run locally and do not exhibit malicious intent.
- [REMOTE_CODE_EXECUTION]: No remote code execution patterns were detected. Downloads are limited to text-based research materials from trusted or standard sources like YouTube.
- [CREDENTIALS_UNSAFE]: No hardcoded credentials or instructions to access sensitive credential files were found.
Recommendations
- CRITICAL: 2 infected file(s) detected - DO NOT USE
- Contains 4 malicious URL(s) - DO NOT USE
Audit Metadata