elevenlabs-stt

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references installation instructions and additional functionality modules hosted on the inference-sh GitHub organization.
  • [COMMAND_EXECUTION]: Uses the belt CLI tool to perform transcription, diarization, and forced alignment tasks through shell commands.
  • [DATA_EXFILTRATION]: Audio data from user-provided URLs is sent to the inference.sh platform for transcription. This behavior is the primary intended function of the skill.
  • [PROMPT_INJECTION]: The skill processes external audio files and user-supplied text strings, which constitutes an indirect prompt injection surface.
  • Ingestion points: Input parameters 'audio' (remote URL) and 'text' in SKILL.md.
  • Boundary markers: Absent in the provided CLI command examples.
  • Capability inventory: Subprocess execution through the belt CLI tool.
  • Sanitization: No evidence of input validation or content filtering before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 06:03 PM
Security Audit — agent-trust-hub — elevenlabs-stt