speech-to-text
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references installation instructions and additional skill modules from the inference-sh GitHub repository, which is the official source for the integrated service.
- [COMMAND_EXECUTION]: Orchestrates model inference using the 'belt' command-line interface. Access to this tool is defined within the skill's allowed-tools configuration.
- [PROMPT_INJECTION]: The skill processes untrusted audio data via 'audio_url' parameters. 1. Ingestion point: audio_url in SKILL.md. 2. Boundary markers: Absent in provided examples. 3. Capability inventory: 'belt' CLI tool for transcription and translation. 4. Sanitization: Not specified. This creates a surface for indirect prompt injection where spoken commands in the audio could influence the agent's behavior, which is an inherent risk for this type of functionality.
Audit Metadata