The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external text input to generate audio, which serves as an ingestion point for potentially untrusted data. Given the specialized output format (audio) and the use of a managed remote API, the risk to the host system is negligible.
Ingestion points: The script scripts/tts.js accepts raw text from command-line arguments.
Boundary markers: No explicit delimiters are used to separate user data from control instructions in the input text.
Capability inventory: The script performs local file writes (WAV files) and network requests to Google's API.
Sanitization: Text input is passed directly to the Gemini API without local filtering or character escaping.
[EXTERNAL_DOWNLOADS]: The skill communicates with Google's Gemini API endpoint via the @google/genai package. These network operations are limited to a well-known service provider and are required for the skill's primary functionality.

gemini-tts