text-to-speech

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches installation scripts and documentation from official inference.sh and Inworld AI repositories and domains.
  • [COMMAND_EXECUTION]: Executes the belt CLI tool for speech synthesis, authentication, and app management. Command access is appropriately restricted to the specific binary in the skill metadata.
  • [PROMPT_INJECTION]: The skill processes untrusted text data for audio synthesis.
  • Ingestion points: User-provided text is passed to the belt CLI via the --input JSON argument.
  • Boundary markers: Uses structured JSON formatting to delineate user text from command arguments.
  • Capability inventory: Limited to executing the belt command as defined in the allowed-tools section.
  • Sanitization: Relies on the security controls and validation implemented by the inference.sh CLI and backend API.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 06:32 PM
Security Audit — agent-trust-hub — text-to-speech