The Agent Skills Directory

[COMMAND_EXECUTION] (SAFE): The skill's main script (scripts/main.py) uses the Whisper library and the standard click CLI framework to process files. It does not invoke arbitrary system shells or execute untrusted code. Instructions in SKILL.md for installing ffmpeg are standard system requirements for audio processing.\n- [EXTERNAL_DOWNLOADS] (SAFE): Dependencies listed in scripts/requirements.txt (openai-whisper, torch, ffmpeg-python, click) are well-known and reputable. While openai-whisper downloads model weights from OpenAI's infrastructure on first use, this is expected behavior for its primary purpose. Under [TRUST-SCOPE-RULE], these downloads from trusted sources (OpenAI, Meta) are considered safe.\n- [DATA_EXFILTRATION] (SAFE): The script processes audio files locally and writes output to the local filesystem. No network requests to unauthorized or suspicious domains were detected in the code.\n- [PROMPT_INJECTION] (SAFE): No evidence of instructions intended to override agent behavior, bypass safety filters, or extract system prompts was found in the documentation or metadata.\n- [DATA_EXPOSURE] (SAFE): No sensitive file paths, credentials, or API keys are accessed or hardcoded within the skill files.

whisper-transcription