The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes system commands to play generated audio files across different platforms.\n
Evidence: The play_audio function in assets/minimax_tts.py calls subprocess.run to invoke system utilities like afplay, aplay, mpv, or ffplay.\n
Context: This execution is a primary feature of the skill. It is implemented securely by passing arguments as a list rather than a single string, effectively preventing shell injection vulnerabilities.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it processes external API data and user-provided text.\n
Ingestion points: Data enters the agent context through the list_voices function (retrieving metadata from the MiniMax API) and the text_to_audio and voice_design functions (handling user-provided text and prompts).\n
Boundary markers: There are no explicit boundary markers or instructions to the agent to disregard potential commands hidden within the processed voice metadata or input text.\n
Capability inventory: The skill has the capability to make network requests, write files to the local system, and execute system commands for audio playback.\n
Sanitization: No sanitization or validation of the input strings or API response content is implemented before processing.

tts-skill