transcribe
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands to run the
whispertranscription tool and the macOSopenutility. User-supplied file paths are properly quoted in the bash script to prevent command injection. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes transcribed text from external media files and instructs the agent to review and 'clean up' the content.
- Ingestion points: Transcribed text generated by the
whispercommand inSKILL.md. - Boundary markers: None; the agent reviews the transcript without specific delimiters or instructions to ignore embedded commands.
- Capability inventory: Shell command execution for transcription and file discovery in
SKILL.md. - Sanitization: None; the raw transcript output is processed directly by the agent.
Audit Metadata