voice-assistant
Pass
Audited by Gen Agent Trust Hub on Jun 26, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes several system commands to process audio, including "ffmpeg" for media conversion, "whisper" for transcription, and "curl" for API requests. It also includes commands for local audio playback ("afplay" and "aplay").
- [EXTERNAL_DOWNLOADS]: The skill makes network requests to the official Mistral AI API ("api.mistral.ai") to perform text-to-speech tasks. This is a well-known service and the implementation uses environment variables for secure credential handling.
- [PROMPT_INJECTION]: The skill has a potential surface for indirect prompt injection because it transcribes external audio content which is intended to be processed by an LLM.
- Ingestion points: User-provided audio files are processed by the "whisper" tool in "SKILL.md".
- Boundary markers: No specific delimiters or instructions to ignore embedded commands are documented when passing transcription results to the LLM.
- Capability inventory: The skill uses subprocess execution for audio processing and network operations across the pipeline in "SKILL.md".
- Sanitization: No text sanitization or filtering of the transcribed output is performed before it is suggested for LLM processing.
Audit Metadata