groq-stt
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMNO_CODEDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- NO_CODE (MEDIUM): The script 'scripts/transcribe.mjs' referenced in the documentation is missing from the provided files. Without the source code, specific behaviors such as arbitrary command execution or unsafe file handling cannot be ruled out.
- Data Exposure & Exfiltration (LOW): The documentation indicates network communication with the Groq API ('api.groq.com') for transcription services. While functional, this domain is not within the defined trusted external sources. There is no evidence in the documentation of sensitive local file access being combined with this network activity.
- Indirect Prompt Injection (MEDIUM): The skill creates a vulnerability surface by processing untrusted audio data. Evidence: 1. Ingestion points: Audio files provided via file path. 2. Boundary markers: Absent. 3. Capability inventory: Network access (to Groq) and file system write access (to save transcripts). 4. Sanitization: None described. Malicious spoken instructions in an audio file could be transcribed and subsequently executed or obeyed by the agent.
- COMMAND_EXECUTION (LOW): The skill instructions require the execution of a local Node.js script ('node scripts/transcribe.mjs'), which is the standard operational method for this skill but involves runtime execution of code.
Audit Metadata