The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the vendor-owned service hotelist.com and its Model Context Protocol (MCP) endpoint (https://hotelist.com/mcp) for structured hotel discovery. This represents a legitimate integration between the skill and the author's own infrastructure.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill processes external, untrusted content from travel platforms, Reddit, and personal blogs using web_search and browser-based tools during its discovery and complaint-mining phases.
Boundary markers: The instructions implement a logical boundary by requiring 'triangulation' across multiple independent sources and specific 'complaint mining' terms, which prevents the agent from over-relying on any single external input.
Capability inventory: The agent's capabilities include web searching, text extraction, and visual analysis of photos, with no access to sensitive system files or local command execution.
Sanitization: The skill prioritizes the extraction of operational data (pricing, amenities, capacity) over free-form text interpretation, naturally limiting the impact of any instructions embedded within hotel reviews.

hotel-hunting