elevenlabs-stt

Pass

Audited by Gen Agent Trust Hub on Aug 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to install the belt-sh/cli and references a remote installation script from the inference-sh GitHub repository. These components are necessary for the skill's documented functionality.\n- [COMMAND_EXECUTION]: The skill uses the belt command-line tool via Bash to perform audio transcription and forced alignment. Access is restricted using the allowed-tools field to only the belt command, following the principle of least privilege.\n- [PROMPT_INJECTION]: The skill processes untrusted audio files from external URLs and user-supplied text for alignment, which constitutes an indirect prompt injection surface.\n
  • Ingestion points: Audio URLs and alignment text provided in the --input JSON payload for the belt tool in SKILL.md.\n
  • Boundary markers: The skill does not define specific delimiters or instructions for the agent to ignore potential commands embedded within the transcribed audio output.\n
  • Capability inventory: The skill is permitted to execute the belt CLI tool via Bash to perform transcription and alignment tasks.\n
  • Sanitization: Transcribed text is passed back to the agent context without explicit sanitization or filtering of the content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 17, 2026, 04:09 PM
Security Audit — agent-trust-hub — elevenlabs-stt