speech-to-text

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references external documentation and installation guides hosted on the vendor's official GitHub repository (inference-sh/skills). It also mentions the installation of a related CLI tool via npm.
  • [COMMAND_EXECUTION]: The skill provides numerous examples of shell commands used to interact with the transcription API. These commands utilize the belt CLI tool, which is explicitly permitted in the skill's configuration (allowed-tools: Bash(belt *)).
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external audio files via URLs for transcription. As this involves processing untrusted third-party content, it theoretically provides a surface for indirect prompt injection if an attacker embeds spoken instructions in the audio; however, this is a standard risk for all transcription-based tools and no specific vulnerabilities were identified.
  • Ingestion points: audio_url, video_url
  • Boundary markers: None specified in instructions
  • Capability inventory: Shell execution (via belt CLI)
  • Sanitization: Dependent on the platform's handling of transcription output.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 09:33 AM
Security Audit — agent-trust-hub — speech-to-text