The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the openai-whisper tool via the Homebrew package manager and downloads model weights from OpenAI's servers to ~/.cache/whisper upon initial execution. These actions involve official tools and well-known services.
[PROMPT_INJECTION]: The skill creates an attack surface for indirect prompt injection by processing external audio data that may contain malicious instructions intended for the agent once transcribed.
Ingestion points: SKILL.md (audio file paths provided to whisper command).
Boundary markers: Absent; there are no instructions to delimit or ignore content within the transcribed text.
Capability inventory: whisper CLI execution, file system reads (input audio), and file system writes (output transcriptions).
Sanitization: Absent; the transcribed text is not filtered or sanitized before being returned to the agent context.

openai-whisper