song-ad
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run local shell commands using 'ffmpeg', 'ffprobe', and 'curl'. It uses variables for timestamps, filenames, and URLs derived from tool outputs. This creates a risk of command injection if these variables contain malicious shell characters.
- [EXTERNAL_DOWNLOADS]: Fetches media assets from the vendor's platform using 'curl' to facilitate the local editing workflow.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user briefs and web research to generate prompts for lyrics, images, and videos. This is an attack surface where embedded instructions could influence the agent.
- Ingestion points: User-provided product descriptions, photos, and web research data.
- Boundary markers: Not explicitly defined in the instructions for prompt construction.
- Capability inventory: Local file system access, shell command execution (ffmpeg, curl), and network uploads/downloads.
- Sanitization: The skill does not mention sanitizing data from external sources before use in prompts or shell commands.
Audit Metadata