tensorslab-video

Pass

Audited by Gen Agent Trust Hub on Mar 2, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface.\n
  • Ingestion points: User-provided prompts and image paths in scripts/tensorslab_video.py are used to construct API requests.\n
  • Boundary markers: The skill does not use delimiters to isolate user-supplied prompt text from internal instructions during API communication.\n
  • Capability inventory: The Python script can make network requests and write generated video files to the local file system (~/tensorslab_output).\n
  • Sanitization: No sanitization or validation logic is applied to the input parameters before they are transmitted to the external API.\n- [EXTERNAL_DOWNLOADS]: Downloads generated video files from the TensorsLab infrastructure.\n
  • Evidence: The download_video function in scripts/tensorslab_video.py fetches data from URLs provided in the API task status response.\n
  • Source: These downloads originate from the vendor's own domains (tensorslab.com).\n- [COMMAND_EXECUTION]: Executes a local script to process video generation requests.\n
  • Evidence: The skill's primary workflow involves the agent running python scripts/tensorslab_video.py with various command-line arguments.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 2, 2026, 02:51 AM