claude-video-shorts

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires installing several well-known Python packages including whisperx, scenedetect, google-genai, mediapipe, and opencv-python. These are established tools for transcription, scene detection, and computer vision.\n- [COMMAND_EXECUTION]: The pipeline relies on the execution of local Bash scripts (scripts/screen_shorts_pipeline.sh, scripts/shorts_pipeline.sh, and scripts/preflight.sh) to coordinate FFmpeg operations and AI analysis.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted video files and uses VLM analysis and transcription to influence its output. \n
  • Ingestion points: External video files processed via the $INPUT argument.\n
  • Boundary markers: No explicit delimiters or instructions to ignore embedded content are defined in the analyzed instructions.\n
  • Capability inventory: File reading/writing and Bash command execution capabilities are available to the skill pipeline.\n
  • Sanitization: No specific sanitization of the VLM-generated visual descriptions or transcription data is documented before these outputs are used to influence clip selection and scoring.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 06:38 PM
Security Audit — agent-trust-hub — claude-video-shorts