The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/generate_indextts2_narration.py executes ffmpeg, ffprobe, git, and uv using argument lists. This method is secure against shell injection attacks.
[EXTERNAL_DOWNLOADS]: The helper assets/minimax_tts.py performs POST requests to the MiniMax API (api.minimax.io) to generate audio. This is the intended behavior for the 'Public Article Lane' described in the skill instructions.
[PROMPT_INJECTION]: The skill processes untrusted text data from JSONL segment contracts and a pronunciation lexicon (references/pronunciation-lexicon.json). While it lacks explicit prompt boundary markers for the TTS engine, it includes a validate_pronunciations mechanism to filter forbidden terms. The capabilities associated with this ingestion include file system writes and subprocess execution of media tools.

tts-skill