competitor-ad-scrape
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill uses local CLI tools including
jq,curl,ffmpeg,ffprobe, and a local build ofwhisper.cppfor data extraction and media processing. These operations are within the scope of the skill's documented functionality. - [EXTERNAL_DOWNLOADS]: Downloads video content from Meta's content delivery networks (via
video_hd_urlorvideo_sd_url). These downloads target a well-known service and are necessary for the transcription and analysis features. - [DATA_EXFILTRATION]: Accesses the local
.mcp.jsonand.envfiles to retrieveAIRTABLE_API_KEYandAPIFY_TOKEN. These credentials are used solely to authenticate with the respective official APIs of these well-known services. - [PROMPT_INJECTION]: The skill includes a proactive safety warning at the top of the instructions, correctly identifying that scraped captions and transcripts are untrusted data. It explicitly instructs the agent to ignore any embedded commands within that content.
Audit Metadata