speech-to-text
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references external resources from the inference-sh GitHub organization, specifically installation instructions for the required 'belt' CLI tool at 'https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md'. These are legitimate dependencies for the skill's stated purpose.
- [COMMAND_EXECUTION]: The skill utilizes the 'belt' command-line interface to execute transcription tasks. The 'allowed-tools' frontmatter correctly restricts shell execution to the 'belt' command, following the principle of least privilege.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it processes external audio files via URLs. However, the risk is mitigated by the use of structured JSON objects for passing arguments to the underlying CLI, and the severity is naturally low for this category.
Audit Metadata