whisper-stt
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSNO_CODEPROMPT_INJECTION
Full Analysis
- [SAFE]: No malicious patterns, such as credential theft or obfuscation, were detected in the skill contents.
- [EXTERNAL_DOWNLOADS]: Recommends installing standard libraries
openai-whisperandtorchfrom official, trusted sources for local processing. - [NO_CODE]: The skill package is documentation-based and does not include the referenced Python implementation scripts like
tools/whisper_stt.py. - [PROMPT_INJECTION]: The skill provides an interface for processing untrusted audio data. Evidence: 1. Ingestion:
audio.mp3via thetranscribe()function. 2. Boundary markers: Absent in the provided usage examples. 3. Capability inventory: Transcribed text is passed to downstream AI processing and voice cloning tools. 4. Sanitization: No sanitization or validation of the transcribed output is performed before use.
Audit Metadata