NYC

speech-to-text

Fail

Audited by Gen Agent Trust Hub on Feb 15, 2026

Risk Level: HIGH
Full Analysis
  • [SAFE] (SAFE): No malicious patterns or security risks were identified. The code examples demonstrate high-standard security practices for data privacy and input validation.
  • [INPUT_VALIDATION] (INFO): The AudioValidator class uses the magic library to perform deep MIME type inspection, preventing attacks involving malicious file headers or incorrect extensions.
  • [DATA_SECURITY] (INFO): SecureAudioStorage implements encryption for audio at rest and a secure wipe procedure (zeroing out) before file deletion.
  • [FALSE_POSITIVE] (INFO): The automated flag for 'logger.info' is a false positive. The scanner misidentified a standard Python method call as a malicious domain.
  • [PROMPT_INJECTION] (INFO): The skill exhibits an indirect prompt injection surface as it processes untrusted audio data.
  • Ingestion points: MultiLanguageSTT.transcribe_auto and WebSocketSTT.handle_stream (references/advanced-patterns.md).
  • Boundary markers: None found in prompt templates.
  • Capability inventory: Logic is restricted to transcription and WebSocket transmission; no dangerous execution or exfiltration capabilities are present in the snippets.
  • Sanitization: Hashing is used for content in logs; PII filtering is mentioned in unit tests but the implementation is not provided.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 15, 2026, 10:30 PM