The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches model code and text processing dependencies from external GitHub repositories.
Evidence: Downloads from https://github.com/OpenMOSS/MOSS-TTS-Nano.git and https://github.com/WhizZest/WeTextProcessing.git.
[COMMAND_EXECUTION]: Executes shell commands to set up the Python environment and run model inference.
Evidence: CLI commands such as conda create, pip install -e ., and moss-tts-nano serve.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it synthesizes speech from external, potentially untrusted text.
Ingestion points: The text and text-file parameters in the generate CLI command and the infer Python method found in SKILL.md.
Boundary markers: Absent. No specific delimiters or instructions are used to prevent the agent from interpreting instructions embedded in the input text.
Capability inventory: The skill can write audio files to the local filesystem via soundfile.write, make local HTTP requests via requests.post, and execute system processes via the moss-tts-nano CLI.
Sanitization: Absent. The skill does not mention any validation or sanitization of the input text before processing.

moss-tts-nano-speech