The Agent Skills Directory

Dynamic Script Generation and Execution: The skill operates by having the agent generate a Python script at work/booth_video.py based on provided templates and then executing it via the shell. While this is the core functional mechanism for frame-by-frame rendering, it involves the runtime execution of code synthesized by the agent.
Indirect Prompt Injection Surface: The skill ingests untrusted user data (video descriptions and brand colors) which is interpolated into the logic of the generated rendering script. There are no explicit boundary markers or sanitization steps mentioned in the instructions to filter this input before it influences the script logic. The agent maintains capabilities to write to the filesystem and execute commands to perform the render.
Automated Dependency Management: The agent uses shell commands to install standard Python packages (pillow, imageio, imageio-ffmpeg, numpy) required for the rendering process. These are well-known libraries for media processing and are retrieved from standard registries.
System Resource Access: The script architecture includes a mechanism to probe standard system font directories (e.g., C:/Windows/Fonts, /Library/Fonts) to ensure typography resolves correctly. It also utilizes system temporary directories for storing intermediate frames to avoid issues with cloud-synchronized folders.

booth-loop-video