faceswap
Pass
Audited by Gen Agent Trust Hub on Mar 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Executes
ffmpegandffprobefor video duration detection and trimming, andcurlfor interacting with the service API.\n- [EXTERNAL_DOWNLOADS]: Fetches remote media content from user-provided URLs and uploads processed data toverging.aiAPI endpoints and Cloudflare R2 storage.\n- [PROMPT_INJECTION]: Indirect prompt injection surface detected through the ingestion of external media files.\n - Ingestion points: Local and remote video and image files specified in command arguments (e.g.,
--video,--face) are processed by the skill.\n - Boundary markers: No explicit boundary markers or 'ignore' instructions are present to mitigate potential instructions embedded in media metadata or content; the skill relies on remote service processing.\n
- Capability inventory: Includes subprocess execution (
ffmpeg,ffprobe), network communication (curl), and local filesystem access (read/write in/tmp/verging-faceswap/).\n - Sanitization: No explicit validation or sanitization of media metadata or binary content is performed before local processing or upload.
Audit Metadata