The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions involve capturing survey responses and potentially uploading them to an external location.
Evidence: The instruction "Ask before any upload" in SKILL.md confirms that the skill has a data transmission workflow.
[INDIRECT_PROMPT_INJECTION]: The skill processes responses from a survey, which represents a surface for instructions to be embedded in data that could influence the agent.
Ingestion points: Survey responses processed in SKILL.md.
Boundary markers: Absent; there are no delimiters or warnings to treat survey data as untrusted content.
Capability inventory: The skill can create the ~/.pua/ directory and modify ~/.pua/config.json.
Sanitization: Absent; the instructions do not require any filtering or escaping of the survey responses before they are saved to the configuration file.

pua-survey