openai-whisper
Pass
Audited by Gen Agent Trust Hub on Apr 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Instructs the agent to execute the
whispercommand line tool to transcribe or translate local audio files. - [EXTERNAL_DOWNLOADS]: Recommends installing the
openai-whisperpackage via the Homebrew package manager and downloads pre-trained model files from OpenAI's infrastructure to the local directory~/.cache/whisperupon first execution. - [PROMPT_INJECTION]: Identifies an indirect prompt injection surface where the agent processes external data (audio) that could contain hidden instructions or malicious content.
- Ingestion points: Reads audio files from user-specified local paths (e.g.,
audio.mp3,audio.m4a). - Boundary markers: Absent in the provided command templates.
- Capability inventory: Execution of the
whisperCLI with parameters for transcription and translation. - Sanitization: No validation or sanitization of the input audio content or the resulting transcription text is described.
Audit Metadata