beat-sync-reel
Pass
Audited by Gen Agent Trust Hub on Jun 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of the
Bashtool to execute system commands includingffmpegfor video encoding,yt-dlpfor audio extraction, andcurlfor web scraping. These commands are necessary for the media processing pipeline described. - [EXTERNAL_DOWNLOADS]: The skill downloads audio content from various third-party platforms (Instagram, TikTok, YouTube) and scrapes images from e-commerce sites. These operations are driven by user-provided URLs and are consistent with the skill's intended functionality.
- [PROMPT_INJECTION]: The skill possesses a potential surface for indirect prompt injection as it ingests untrusted data from external product pages (such as Shopify JSON or HTML metadata) and processes it into text overlays on the final video.
- Ingestion points: Product page URLs (Shopify JSON or HTML) and audio metadata (SKILL.md).
- Boundary markers: None identified in the provided instructions.
- Capability inventory: Subprocess calls via
Bash(ffmpeg, yt-dlp, curl), file-write operations, andWebSearchcapability. - Sanitization: No explicit sanitization or validation of the extracted product information (brand name, product name, price) is mentioned before it is passed to the Pillow-based image generation script.
Audit Metadata