hotel-all-in-one

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.90). These URLs include direct downloads of platform binaries (.exe and standalone rgh binaries) hosted in a little-known GitHub repo plus a custom auth domain (rollinggo.store) referenced by the installer/skill — a common pattern for distributing unsigned executables and OAuth redirects that is high risk without additional trust/verification.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill contains explicit runtime fetches of remote content: SKILL.md directs the agent to silently GET https://raw.githubusercontent.com/RollingGo-AI/rollinggo-hotel-skill-CN/main/skills/{name}/SKILL.md to check versions (which changes agent output), and scripts/install.py performs runtime requests to the GitHub Releases API and downloads executables from https://api.github.com/repos/RollingGo-AI/oauth-hotel-cli/releases/latest and related https://github.com/RollingGo-AI/oauth-hotel-cli/.../download/... URLs to fetch binaries the skill relies on.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). 该 Skill 明确包含专门的支付/下单相关命令：使用 rgh price-confirm 锁定价格、使用 rgh book 创建订单，并从结果中提取 alipayUrl 或其他支付链接返回给用户。虽然 Agent 本身不直接“代付”，但这些命令是专门用于发起真实消费（生成待支付订单并提供支付链接）的明确金融执行操作，因此属于直接金融执行能力范畴。