moss-tts-nano-speech
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of the MOSS-TTS-Nano repository from GitHub and fetches model weights from Hugging Face or ModelScope mirrors.- [COMMAND_EXECUTION]: Provides instructions for installing dependencies using pip and conda, including installation from a requirements file and local editable installation of the package.- [DYNAMIC_EXECUTION]: Utilizes a dynamic import pattern (import('base64')) within a Python code snippet to encode audio data for an API request.- [INDIRECT_PROMPT_INJECTION]: The skill accepts arbitrary text input for speech synthesis via CLI arguments and API calls.
- Ingestion points: Ingests untrusted data through the text parameter in Python functions and the --text/--text-file flags in the CLI entrypoints.
- Boundary markers: No specific delimiters or safety instructions are provided to separate user-supplied input from the synthesis instructions.
- Capability inventory: The skill utilizes file system access for saving generated audio and network capabilities for communicating with a local server.
- Sanitization: No input sanitization or filtering of the synthesis text is implemented in the provided code examples.
Audit Metadata