tts
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to construct and execute shell commands where user-provided text is directly interpolated as an argument (e.g.,
pocket-tts generate --text "..."). This pattern creates a command injection risk if the input text contains shell metacharacters like backticks or semicolons and is not properly sanitized or escaped. - [EXTERNAL_DOWNLOADS]: The skill references downloading model weights from Hugging Face's official repository for Kyutai Pocket TTS. This is a standard operation required for functionality and targets a well-known platform.
- [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface when processing untrusted user data:
- Ingestion points: User-supplied text intended for speech generation in
SKILL.md. - Boundary markers: The instructions use double quotes as delimiters for the text argument in shell commands.
- Capability inventory: Shell command execution for TTS generation, file management (
mktemp,rm), and audio playback (afplay). - Sanitization: No explicit sanitization or validation of the input text is requested beyond the use of quote delimiters.
Audit Metadata