pricewin-hotel-search

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill comprises instructional markdown and does not contain any executable scripts, binary files, or installation commands. It serves as a guidance layer for an external MCP server.\n- [DATA_EXFILTRATION]: The skill transmits travel search parameters (city, check-in/out dates, guest count) to the PriceWin service. This is a necessary and transparent part of its search functionality. No attempts were found to access or exfiltrate sensitive local data like environment variables, SSH keys, or cloud credentials.\n- [PROMPT_INJECTION]: The skill instructions focus on controlling tool flow (polling logic) and formatting search results. There are no patterns indicative of attempts to bypass safety filters or override the agent's core instructions.\n- [COMMAND_EXECUTION]: The skill directs the agent to perform string manipulation to append dates to booking URLs for Agoda and Booking.com. This is handled as text processing rather than shell command execution, and targets reputable, well-known domains.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 06:02 AM
Security Audit — agent-trust-hub — pricewin-hotel-search