The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides documented examples of using curl, jq, and ffmpeg to interact with its synthesis API and manipulate audio files, which are standard utilities for this use case.
[DATA_EXFILTRATION]: Network communication is directed to the vendor's synthesis endpoint ($API_URL). Authentication is managed via context-provided variables ($AUTH_TOKEN), avoiding hardcoded credentials.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it processes untrusted text for audio synthesis.
Ingestion points: The text and instructions parameters in SKILL.md accept external content.
Boundary markers: No explicit delimiters are used to wrap the input text.
Capability inventory: The skill uses curl to transmit data to the external API.
Sanitization: No specific sanitization or filtering of the input text is described.

text-to-speech