The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions utilize curl for API interactions and chmod 600 to ensure the local API key file is restricted to the current user.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text strings to generate audio, creating a potential vector for indirect instructions if the input is not sanitized by the underlying model.
Ingestion points: The prompt and audio_url fields in the POST request to the Pixazo gateway (found in SKILL.md).
Boundary markers: None identified in the prompt templates.
Capability inventory: File system read/write access (~/.pixazo/api-key), network communication (gateway.pixazo.ai), and shell command execution (curl, chmod).
Sanitization: No explicit sanitization or validation of the input text is performed before it is sent to the API.

lux-tts