The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts tts.py and mouth.py use subprocess.run to invoke ffmpeg for audio conversion, tempo scaling, and waveform measurement. These calls use specific argument lists rather than shell strings, which is a secure practice for command execution.
[EXTERNAL_DOWNLOADS]: The skill fetches audio and alignment data from ElevenLabs (api.elevenlabs.io), which is a well-known AI audio service. The automated scanner alerts for 'Remote Code Execution' are heuristic detections of this data-to-subprocess pipeline, which is the intended and standard behavior for a TTS utility.
[SAFE]: Credential management is handled securely by reading from the environment or a .env file. The documentation explicitly advises against insecure practices like pasting keys into chat logs or public repositories.
[SAFE]: All external references and dependencies, including the use of the uv package manager and numpy library, are standard for Python-based media processing tools.

voiceover