oma-voice

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection during the processing of external content.
  • Ingestion points: The skill reads local audio files for transcription (STT) and accepts text strings for speech generation (TTS).
  • Boundary markers: There are no specific delimiters or instructions defined to prevent the agent from following commands that might be embedded in transcribed text or input strings.
  • Capability inventory: The skill allows writing files to the local filesystem and executing MCP tools for audio manipulation.
  • Sanitization: While the skill enforces character and duration limits, it does not perform semantic validation or sanitization of the content being processed.
  • [EXTERNAL_DOWNLOADS]: The instructions include a reference to the Voicebox desktop application's official GitHub repository to guide the user through the installation process.
  • [COMMAND_EXECUTION]: The skill performs operations by calling tools on a locally hosted Voicebox MCP server and making HTTP requests to a loopback address (127.0.0.1) for health checks and status monitoring.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 12:59 PM
Security Audit — agent-trust-hub — oma-voice