openai-whisper
Pass
Audited by Gen Agent Trust Hub on Mar 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill initiates the installation of the
openai-whisperpackage via the Homebrew package manager. - [EXTERNAL_DOWNLOADS]: The tool is designed to download machine learning models from OpenAI's servers to the local directory
~/.cache/whisperduring its initial run. - [COMMAND_EXECUTION]: The skill provides command patterns for executing the
whisperCLI tool to perform audio transcription and translation on local files. - [PROMPT_INJECTION]: The skill processes untrusted audio data which serves as a surface for indirect prompt injection, as the resulting transcription could contain instructions that influence the agent's behavior.
- Ingestion points: Audio files provided to the whisper CLI (SKILL.md).
- Boundary markers: None; the skill does not specify markers to delimit the transcribed content.
- Capability inventory: Local CLI execution of the whisper binary.
- Sanitization: None; the transcription process operates on raw audio data without filtering.
Audit Metadata