The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/batch_generate.py and scripts/setup_env.py scripts utilize the subprocess.run function to manage environment setup and coordinate the batch generation of videos. These calls are executed using argument lists rather than shell strings, which is a security best practice that prevents command injection vulnerabilities.
[EXTERNAL_DOWNLOADS]: During the environment preparation phase in scripts/setup_env.py, the skill downloads and installs several well-known and standard Python libraries (opencv-python, numpy, and av) from the official Python Package Index (PyPI). This is expected behavior for a tool focused on image processing and video encoding.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-provided images, which serves as a potential attack surface for indirect prompt injection, although the risk is minimized by the skill's specific focus on visual data transformation.
Ingestion points: The skill accepts images via the image_path argument in generate_whiteboard.py and the --images array in batch_generate.py.
Boundary markers: None are present to distinguish between image data and potential instructions.
Capability inventory: The skill has the capability to write MP4 files to the local file system and execute other internal Python scripts.
Sanitization: The skill resizes and processes input images at the pixel level, which reduces the likelihood of an agent interpreting hidden text instructions inside the visual data.

whiteboard-animation