vox-video-director
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill extensively uses the
subprocessmodule to invoke external binaries, includingffmpeg,ffprobe, and the Aliyun Bailian CLI (bl). These tools are used for video transcoding, metadata extraction, and interacting with Alibaba Cloud's model services, which is necessary for the skill's intended functionality. - [EXTERNAL_DOWNLOADS]: The skill connects to
api.atlascloud.aiand Alibaba Cloud infrastructure to upload source media and download AI-generated assets (images, video clips, and synthesized speech). These network operations are standard for a cloud-integrated media skill. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting user-provided topics and media metadata and interpolating them into complex prompt templates for image and video generation models.
- Ingestion points: User-provided topics in
beats.json, source videos in A-roll mode, and anchor photos in C-roll mode. - Boundary markers: The skill does not currently use explicit delimiters or 'ignore' instructions to isolate user content within its generated prompts.
- Capability inventory: The skill possesses the ability to write to the local file system, perform network API calls, and execute shell commands through
ffmpegandbl. - Sanitization: There is minimal sanitization of user-provided strings before they are inserted into the prompt construction logic.
Audit Metadata