speech-to-text
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted audio/video files to generate transcripts. If these transcripts contain malicious commands, the agent might execute them if not properly isolated. \n
- Ingestion points: External audio/video files are ingested via
client.speech_to_text.convert. \n - Boundary markers: Absent. The output text is returned to the agent without specific delimiters or instructions to treat it as untrusted data. \n
- Capability inventory: The skill performs network requests to
api.elevenlabs.io. \n - Sanitization: No filtering is performed on the transcription results. \n- [EXTERNAL_DOWNLOADS]: The skill utilizes several third-party packages for its operations. \n
- Python dependencies include
elevenlabs,python-dotenv, andpydub. \n - Node.js dependencies include
@elevenlabs/elevenlabs-js,@elevenlabs/client,@elevenlabs/react, anddotenv. \n - These packages are recognized as official or standard tools for the described use case.
Audit Metadata