The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external text data (such as tokens generated by an LLM) and convert it into audio, creating an attack surface for indirect prompt injection.
Ingestion points: Data enters the context via the text parameter in the client.speak.v1.audio.generate method (REST) and the SpeakV1Text object in the conn.send_text method (WebSocket).
Boundary markers: No specific delimiters or instructions to ignore embedded commands are demonstrated in the provided code snippets.
Capability inventory: The skill possesses the capability to perform network requests to the Deepgram API and write binary audio data to the local file system (e.g., output.raw).
Sanitization: The documentation does not demonstrate explicit input sanitization or validation of the text string before it is sent to the synthesis engine.
[COMMAND_EXECUTION]: The skill provides a command-line instruction to expand the agent's capabilities using the npx package runner.
Evidence: npx skills add deepgram/skills in the 'Central product skills' section.
Analysis: The command is used to install additional skills from the official vendor's repository (deepgram/skills), which is an expected action within the context of using the vendor's SDK tools.

deepgram-python-text-to-speech