██████╗ ██████╗ ██╗ ██╗ ██╗███╗ ██╗ ██████╗ ██████╗ ██████╗
██╔══██╗██╔═══██╗██║ ██║ ██║████╗ ██║██╔════╝ ██╔════╝ ██╔═══██╗
██████╔╝██║ ██║██║ ██║ ██║██╔██╗ ██║██║ ███╗██║ ███╗██║ ██║
██╔══██╗██║ ██║██║ ██║ ██║██║╚██╗██║██║ ██║██║ ██║██║ ██║
██║ ██║╚██████╔╝███████╗███████╗██║██║ ╚████║╚██████╔╝╚██████╔╝╚██████╔╝
╚═╝ ╚═╝ ╚═════╝ ╚══════╝╚══════╝╚═╝╚═╝ ╚═══╝ ╚═════╝ ╚═════╝ ╚═════╝
██╗ ██╗ ██████╗ ████████╗███████╗██╗
██║ ██║██╔═══██╗╚══██╔══╝██╔════╝██║
███████║██║ ██║ ██║ █████╗ ██║
██╔══██║██║ ██║ ██║ ██╔══╝ ██║
██║ ██║╚██████╔╝ ██║ ███████╗███████╗
╚═╝ ╚═╝ ╚═════╝ ╚═╝ ╚══════╝╚══════╝
──●────●────●────●────●────●──
AI Skills for Claude • Cursor • Copilot
Search → Detail → Price Confirm → Book → Orders
Official AI skills for the RollingGo Hotel CLI (@rollinggo/hotel). They teach agents the correct booking sequence, OAuth 2.1 PKCE login flow, and how to safely trigger end-to-end hotel transactions from the terminal. Works with the skills CLI (Cursor, Claude Code, Codex, Windsurf, Copilot, 40+ agents).
Important: This CLI acts as a bridge to real-world bookings. AI agents are strictly instructed to NEVER place an order without explicit price confirmation from the user.
For AI agents: When the user asks to book a hotel, search for rooms, or check their orders, use the rgh CLI commands sequentially. Never leak hotelId or referenceNo directly to the user; format results into user-friendly Markdown.
Works with Cursor, Claude Code, Codex, Windsurf, Copilot, Google Antigravity, and 40+ agents:
# Install interactively (Select skills from menu)
npx skills add https://github.com/RollingGo-AI/hotel-skill
# Or install a specific skill directly (e.g., hotel-core)
npx skills add https://github.com/RollingGo-AI/hotel-skill --skill hotel-coreThe CLI auto-detects the installed agent. To target one explicitly, pass --agent:
npx skills add https://github.com/RollingGo-AI/hotel-skill --agent cursorClone this repo and copy the skills/hotel-core/ folder into the appropriate directory for your agent:
| Agent | Skill Directory |
|---|---|
| Claude Code | ~/.claude/skills/ |
| Cursor | ~/.cursor/skills/ |
| OpenAI Codex | ~/.codex/skills/ |
| Windsurf | ~/.windsurf/skills/ |
| Google Antigravity | ~/.gemini/antigravity/skills/ |
| Command | Description |
|---|---|
rgh login |
OAuth 2.1 PKCE Authorization |
rgh search-hotels |
Search hotels by location, dates, tags |
rgh hotel-detail |
View room types, live prices, cancellation policies |
rgh price-confirm |
Lock a price and generate a booking referenceNo |
rgh book |
Finalize order with guest info, returns Alipay link |
rgh orders |
View historical orders and their payment status |
Canonical order sequence to suggest or generate:
# 1. Start with checking login status
rgh whoami
# -> If not logged in, prompt user to run `rgh login`
# 2. Search for hotels (e.g. Shanghai next week)
rgh search-hotels --place "Shanghai" --check-in-date 2026-06-20
# 3. Present options to user and get their choice. Then fetch details:
rgh hotel-detail --hotel-id 12345 --check-in-date 2026-06-20
# 4. Once user picks a room and rate plan, lock the price (Requires user confirmation!)
rgh price-confirm --hotel-id 12345 --rate-plan-id "RP01" --rooms 1 --check-in-date 2026-06-20 --check-out-date 2026-06-21 --adults 2
# 5. After the user approves the locked price and provides their Name/Email:
rgh book --reference-no "REF9999" --first-name "San" --last-name "Zhang" --email "zhangsan@example.com"MODERATE — Can trigger real-world hotel reservations. The skill enforces a strict two-step confirmation gate before making any book calls.
为 RollingGo 酒店命令行工具 (@rollinggo/hotel) 量身定制的官方 AI 技能库。它能够教会各类 AI 代理正确的酒店预订流程、OAuth 2.1 PKCE 登录流，以及如何安全地在终端内触发真实的酒店交易。兼容 skills CLI 规范，支持 Cursor、Claude Code、Codex、Windsurf、Copilot 等 40 多种主流大模型代理。
安全警告：本工具直接连接真实交易系统。技能库已向 AI 代理下达严格的“死命令”：在未经用户明确确认价格和支付意愿前，绝不允许私自发起任何订单。
给 AI 代理的提示： 当用户要求预订酒店、查询空房或查看订单时，请按顺序使用 rgh 命令。绝对不要向用户泄露原始的 hotelId 或 referenceNo 等系统内部 ID，请将结果格式化为美观的 Markdown 卡片。
自动兼容 Cursor, Claude Code, Windsurf 等 40+ 种代理：
# 交互式安装 (推荐，可从菜单中自由勾选不同能力模块)
npx skills add https://github.com/RollingGo-AI/hotel-skill
# 或直接定向安装特定核心模块 (跳过交互选单)
npx skills add https://github.com/RollingGo-AI/hotel-skill --skill hotel-coreCLI 会自动检测您系统上安装的 Agent。您也可以通过 --agent 强制指定目标环境：
npx skills add https://github.com/RollingGo-AI/hotel-skill --agent cursor克隆本仓库，将 skills/hotel-core/ 文件夹直接放入您 Agent 对应的技能目录即可：
| 代理工具 | 本地技能存放目录 |
|---|---|
| Claude Code | ~/.claude/skills/ |
| Cursor | ~/.cursor/skills/ |
| OpenAI Codex | ~/.codex/skills/ |
| Windsurf | ~/.windsurf/skills/ |
| Google Antigravity | ~/.gemini/antigravity/skills/ |
| 指令 | 业务描述 |
|---|---|
rgh login |
OAuth 2.1 PKCE 免密安全授权 |
rgh search-hotels |
支持按地标、日期、标签多维度检索酒店 |
rgh hotel-detail |
实时拉取酒店房型、最新报价与退改政策 |
rgh price-confirm |
锁定房间价格并生成交易参考号 referenceNo |
rgh book |
提交预订入住人信息，返回真实的支付宝付款链接 |
rgh orders |
查阅历史订单及最新支付状态 |
要求 AI 建议或自动执行的规范代码流：
# 1. 首先检查用户登录状态
rgh whoami
# -> 如果未登录，引导用户先执行 `rgh login`
# 2. 检索意向酒店（例如：下周的上海酒店）
rgh search-hotels --place "Shanghai" --check-in-date 2026-06-20
# 3. 将搜索结果用 Markdown 呈现给用户。待用户挑选后，获取详情：
rgh hotel-detail --hotel-id 12345 --check-in-date 2026-06-20
# 4. 用户确定房型后，进行锁价预订 (此步为高风险，必须要求用户确认！)
rgh price-confirm --hotel-id 12345 --rate-plan-id "RP01" --rooms 1 --check-in-date 2026-06-20 --check-out-date 2026-06-21 --adults 2
# 5. 在用户同意上述锁定的价格并提供姓名邮箱后，正式下单：
rgh book --reference-no "REF9999" --first-name "San" --last-name "Zhang" --email "zhangsan@example.com"中等风险 (MODERATE) — 此技能可触发真实的现实世界酒店客房交易。已通过严格的两步确认机制拦截未经授权的 book 调用。
RollingGo-Skills/
├── README.md # Main orchestrator and SEO entry / 主入口与 SEO 文案
├── AGENTS.md # Directives for AI web crawlers / 针对 AI 爬虫的全局指令
├── .github/
│ └── copilot-instructions.md
└── skills/
├── llms.txt # AI Index for search engines / 大模型全局搜索索引
└── hotel-core/ # Core hotel booking skill / 核心酒店预订功能包
├── SKILL.md
└── references/
└── cli-params.md
MIT