The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts scripts/tts_measure.py and scripts/stitch_mux.py call external utilities including ffmpeg, ffprobe, and volc-tts. These calls are implemented using Python's subprocess.run with argument lists, which prevents shell command injection by ensuring that inputs are not interpreted as shell commands.
[PROMPT_INJECTION]: The skill ingests narration text from a user-supplied plan.md file. Analysis of this indirect input surface shows it is handled safely; the extracted text is passed as a literal string to the TTS engine via subprocess calls, preventing any embedded instructions from influencing the agent's logic or the system's execution.

manim-video-with-tts