The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses ffmpeg and ffprobe to calculate durations and merge audio files. The commands are constructed using lists in tts.py, which is a secure practice to prevent shell injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes script.json which contains untrusted text and identifiers.
Ingestion points: The script_path file is read and parsed as JSON, specifically accessing the sections, id, and lines fields in tts.py.
Boundary markers: No delimiters or ignore instructions are present to distinguish between instructions and data within the script file.
Capability inventory: The skill has file system write access via edge_tts.save() and timing_path.write_text() in tts.py, and performs command execution via ffmpeg.
Sanitization: While ffmpeg calls are structured as lists, the section id identifier is used to construct file paths without validation or escaping, presenting a potential directory traversal surface if a malicious script is processed.
[EXTERNAL_DOWNLOADS]: The skill utilizes the edge-tts library to synthesize audio, which involves communication with external Microsoft TTS servers. This is consistent with the skill's stated purpose and uses a well-known service provider.

tts-from-script