The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs video rendering tasks by invoking the FFmpeg binary through the subprocess.run function in render.py, assets.py, and doctor.py. All command invocations utilize argument lists rather than shell strings and omit the shell=True parameter, effectively neutralizing the risk of shell injection attacks. \n- [SAFE]: Several security-focused utilities are implemented in safety.py. The safe_slug function ensures that user-provided strings are sanitized before being used as filenames, while ensure_within_dir provides a safeguard against directory traversal by verifying that file operations are restricted to the project's root directory. \n- [SAFE]: The system implements a mandatory approval gate for final renders. As evidenced in project.py, the write_approval function generates a SHA256 hash of the preview video, which is later used to verify that the final rendering plan has not been tampered with since the last approved review. \n- [SAFE]: Static analysis identified the use of __import__ in project.py to dynamically load standard library modules hashlib and datetime. Upon review, this usage is benign and localized to standard utility operations, posing no security threat.

youtube-factory-pro