The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection vulnerability surface. The skill processes data from third parties that could contain malicious instructions.
Ingestion points: Untrusted data enters via client.ts through methods such as reservations.listMessages, inquiries.list, and reviews.list.
Boundary markers: Instructions in SKILL.md and rules.md do not define delimiters for external data.
Capability inventory: The agent can execute write actions via client.ts including reservations.sendMessage, calendar.update, and shortcodes.set.
Sanitization: The skill does not implement validation or escaping of the fetched API content.

hospitable-api