skills/zju-real/easel/video-editing/Gen Agent Trust Hub

video-editing

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script skills/shared/scripts/video_ops.py to perform video processing tasks. This script serves as a wrapper for ffmpeg and ffprobe binaries. The instructions explicitly direct the agent to use this script instead of hand-crafting shell commands, which reduces the risk of improper shell syntax or unintended command behavior.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it processes user-supplied text for video overlays and file paths.
  • Ingestion points: User-provided video file paths and text strings provided via the --text argument in the text subcommand.
  • Boundary markers: The skill does not explicitly define delimiters for user-supplied data within its instructions, relying on the wrapper script for parameter handling.
  • Capability inventory: The skill can read and write to the file system (specifically in the outputs/ directory) and execute subprocesses via the video_ops.py script.
  • Sanitization: The documentation indicates that the script handles common error points and parameter defaults, though the specific shell escaping logic for user strings is encapsulated within the script itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 01:22 PM