speech-to-text

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references installation guidelines from the inference-sh GitHub repository and platform documentation from inference.sh. It also provides instructions to install related skill modules using the npx skills add command.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it transcribes untrusted audio data that may contain hidden instructions for the agent.
  • Ingestion points: Processes external audio files via the audio_url parameter in multiple examples within SKILL.md.
  • Boundary markers: There are no instructions or delimiters provided to warn the agent to ignore instructions embedded in the resulting transcription.
  • Capability inventory: The skill uses the Bash tool with the belt CLI to perform audio processing and transcription.
  • Sanitization: No sanitization or validation of the transcribed text is performed before it is returned to the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 06:03 PM
Security Audit — agent-trust-hub — speech-to-text