gemini-viral-shorts
Pass
Audited by Gen Agent Trust Hub on Apr 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill adheres to recommended security practices by managing the Google Gemini API key through environment variables and local configuration files rather than hardcoding credentials.
- [COMMAND_EXECUTION]: Video processing is performed using the yt-dlp library's Python API instead of direct shell execution, which mitigates the risk of command injection through malicious YouTube URLs.
- [EXTERNAL_DOWNLOADS]: Network operations and downloads are restricted to well-known and reputable services, specifically YouTube and the Google Gemini API platform.
- [DATA_EXFILTRATION]: No unauthorized data transmission or sensitive file access was detected. The script correctly cleans up temporary video files from the local filesystem after processing is complete.
- [PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection as it processes untrusted YouTube content to generate text.
- Ingestion points: Video content and transcripts are ingested in scripts/gemini_viral_shorts_post.py.
- Boundary markers: The model prompt does not use delimiters to isolate external content from system instructions.
- Capability inventory: The skill possesses file-write access for temporary storage and network access to Google and YouTube APIs.
- Sanitization: No explicit filtering or sanitization is applied to the ingested external data.
Audit Metadata