The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides patterns for processing user-supplied text to generate audio, which creates a potential surface for indirect prompt injection where instructions hidden in the processed text might attempt to influence the agent's behavior.
Ingestion points: User-provided text is passed into the prompt field within the JSON input for the belt app run command in SKILL.md.
Boundary markers: The examples do not demonstrate the use of explicit delimiters or negative constraints to prevent the underlying model from following instructions contained within the dialogue text.
Capability inventory: The skill utilizes the Bash tool with access restricted to the belt CLI command to interact with external AI inference services.
Sanitization: No explicit input sanitization or validation steps are provided in the instructions for the dialogue text before it is sent to the inference engine.

dialogue-audio