The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from both the user (text to be converted) and external API responses (JSON data from the gateway), creating a potential surface for indirect instruction injection.
Ingestion points: The skill ingests data from external API responses at gateway.pixazo.ai and user-supplied strings for voice synthesis as described in the 'Make the API call' section of SKILL.md.
Boundary markers: There are no explicit delimiters or specific instructions provided to the agent to disregard or escape potential instructions embedded within the processed text or the JSON response body.
Capability inventory: The skill utilizes network capabilities via curl, Python requests, and Node.js fetch to interact with remote endpoints, and includes file system write access to store API credentials.
Sanitization: The instructions do not specify sanitization or validation logic for the content of the API response (e.g., the audio.url field) before it is displayed to the user or acted upon.

qwen-tts