The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python utility scripts, scripts/analyze-audio.py and scripts/measure-motion.py, which execute external binaries such as ffmpeg and demucs via the subprocess module. The scripts utilize argument lists for subprocess.run(), which is a secure practice to prevent shell injection attacks. The demucs path can be influenced by the DEMUCS_BIN environment variable, a standard configuration feature for locating binaries.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and analyzing user-provided external media files (audio and video) and style references. This processing of external data constitutes an attack surface where metadata or content could attempt to influence the agent's behavior. The risk is mitigated by a 'gate model' requiring explicit user approval of artifacts (SSOT, storyboard, execution plan) and the use of deterministic scripts for objective data extraction (BPM, energy levels).
[SAFE]: The skill demonstrates well-architected project management and automation, emphasizing state tracking via PROGRESS.md and immutable 'frozen' execution plans to ensure predictable AI behavior.

klik-remotion-method