avatar-video
Fail
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: CRITICALEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill documentation describes workflows for fetching remote media assets from user-supplied URLs and downloading generated video files from HeyGen's infrastructure (files.heygen.ai, heygen.com). These network operations are standard for a video generation service.
- [COMMAND_EXECUTION]: The skill includes curl command examples for API interaction, demonstrating how to use authentication headers and manage video generation tasks via the command line.
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a proxy for user-supplied scripts and media assets, which are interpolated into API requests for video generation. Ingestion points: Script parameters in references/video-generation.md and remote asset URLs in references/assets.md. Boundary markers: Absent; scripts and URLs are interpolated directly into request objects. Capability inventory: HTTP network requests and local file system writes (fs.writeFileSync) for downloading results. Sanitization: No explicit sanitization or validation of the input script content is shown in the examples, relying on the target platform's internal processing.
Recommendations
- CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
Audit Metadata