video-noise-reduction
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data via the video_urls parameter, which represents a standard indirect injection surface for video-processing tools. 1. Ingestion points: The video_urls field in the API request examples in SKILL.md. 2. Boundary markers: None present in the skill instructions. 3. Capability inventory: The skill is limited to performing network requests to authorized EachLabs API endpoints. 4. Sanitization: No explicit sanitization logic is provided in the skill instructions.
- [SAFE]: The skill follows security best practices by utilizing the $EACHLABS_API_KEY environment variable for authentication rather than hardcoding credentials.
- [SAFE]: All network operations are directed towards official vendor-owned domains, specifically eachlabs.run and eachlabs.ai, posing no risk of unauthorized data exfiltration.
Audit Metadata