text-to-speech

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references installation instructions and resource files hosted on the official inference.sh GitHub repository.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided text for audio synthesis, creating a potential surface for instructions embedded within the text data.
  • Ingestion points: Text strings provided by the user via the --input flag in the belt CLI tool (SKILL.md).
  • Boundary markers: Content is passed within JSON structures, though no internal delimiters are used for the text values themselves.
  • Capability inventory: Limited to shell execution of the belt command via Bash(belt *).
  • Sanitization: Input sanitization and safety filtering are handled by the external inference service provider.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 02:18 PM
Security Audit — agent-trust-hub — text-to-speech