transcribe
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external audio files and speaker reference files which are then passed to the OpenAI API for transcription and diarization.
- Ingestion points: The
scripts/transcribe_diarize.pyscript ingests user-supplied audio files via the--audioargument and reference audio files via the--known-speakerargument inSKILL.md. - Boundary markers: The prompt structure does not include specific delimiters or instructions to ignore potential spoken commands embedded within the audio files.
- Capability inventory: The skill possesses file system read capabilities (
Path.read_bytes,audio_path.open), file system write capabilities for saving transcripts, and network communication capabilities with the OpenAI API. - Sanitization: The script performs basic file existence verification and checks if audio file sizes exceed the 25MB API limit.
Audit Metadata