Setup
If ~/tripadvisor/ does not exist or is empty, read setup.md, explain local storage in plain language, and ask for confirmation before creating files.
When to Use
User wants to interact with Tripadvisor directly: search destinations, compare hotels/restaurants/attractions, inspect reviews, or build trip shortlists using official API workflows and stable web navigation patterns.
Architecture
Memory lives in ~/tripadvisor/. See memory-template.md for setup.
CODEBLOCK0
Quick Reference
| Topic | File |
|---|
| Setup flow | INLINECODE4 |
| Memory schema |
memory-template.md |
| Official API workflows |
api-workflows.md |
| UI navigation playbook |
web-navigation.md |
| Terms and compliance boundaries |
compliance.md |
Core Rules
1. Use official Tripadvisor interfaces only
Use official Tripadvisor API endpoints and normal browser navigation. Never scrape hidden data, bypass access controls, or automate prohibited extraction.
2. Choose mode explicitly at start
Pick one mode per task:
- - API mode: structured lookup and machine-friendly fields
- UI mode: interactive browsing and shortlist guidance
- Hybrid mode: API for discovery, UI for final verification
3. Resolve location_id before deep queries
For API mode, first map user query to a valid
location_id, then fetch detail/review/photo endpoints. Cache successful mappings in
~/tripadvisor/api/location-cache.md.
4. Prefer URL-driven navigation over fragile clicks
In UI mode, rely on stable Tripadvisor URLs when possible (city vertical pages and entity detail URLs) before complex click chains.
5. Handle consent and anti-bot states safely
If cookie dialogs or anti-bot interstitials appear, ask for user confirmation, document the blocker, and continue with API mode or direct URLs. Do not attempt bypass techniques.
6. Produce decision-ready outputs
Always return a ranked shortlist with explicit tradeoffs:
- - fit to user goal
- price/quality signal
- uncertainty or missing data
7. Keep storage minimal and transparent
Store only reusable trip preferences and selected options under
~/tripadvisor/. Confirm first write in a session, avoid sensitive personal data, and never store secrets in logs (always redact API keys).
Common Traps
- - Starting with deep detail calls before resolving
location_id -> API responses become noisy or unusable. - Depending on brittle DOM selectors -> flow breaks when Tripadvisor UI changes.
- Mixing API and UI data without timestamps -> stale comparisons and wrong recommendations.
- Ignoring cookie/consent overlays -> automated clicks fail and create false negatives.
- Treating ratings alone as truth -> misses review recency and recurring complaint patterns.
- Sending unnecessary user data in queries -> avoidable privacy risk and compliance issues.
External Endpoints
| Endpoint | Data Sent | Purpose |
|---|
| INLINECODE14 | destination queries and language scope | resolve INLINECODE15 |
| INLINECODE16 and related location subpaths |
location_id, language, and lightweight filters | fetch place details, reviews, photos, nearby context |
|
https://www.tripadvisor.com/* | standard browser navigation signals and user search terms | inspect public pages and verify shortlist candidates |
No other data is sent externally.
Security & Privacy
Data that leaves your machine:
- - Destination names, optional date windows, and lightweight filters sent to Tripadvisor API or web pages.
Data that stays local:
- - Preferences, shortlist decisions, and request logs in
~/tripadvisor/.
This skill does NOT:
- - Access files outside INLINECODE20
- Store payment or passport data by default
- Use scraping bypasses, CAPTCHA evasion, or anti-bot circumvention
Trust
By using this skill, query data is sent to Tripadvisor services.
Only install if you trust Tripadvisor with the search terms and travel context you provide.
Scope
This skill ONLY:
- - Operates Tripadvisor discovery and comparison workflows
- Uses official API-first calls and policy-safe web navigation
- Produces ranked shortlists with transparent tradeoffs
This skill NEVER:
- - Claim guaranteed booking outcomes
- Present uncertain data as verified facts
- Execute purchases or account actions without explicit user instruction
Related Skills
Install with
clawhub install <slug> if user confirms:
- -
booking — compare accommodation options and total cost breakdowns - INLINECODE23 — manage broader trip planning workflows
- INLINECODE24 — validate route friction and area accessibility on macOS
- INLINECODE25 — improve query iteration and source triangulation
- INLINECODE26 — track trip spending after shortlist decisions
Feedback
- - If useful: INLINECODE27
- Stay updated: INLINECODE28
设置
如果 ~/tripadvisor/ 不存在或为空,请阅读 setup.md,用通俗语言解释本地存储,并在创建文件前请求确认。
使用时机
用户希望直接与 Tripadvisor 交互:搜索目的地、比较酒店/餐厅/景点、查看评论,或使用官方 API 工作流和稳定的网页导航模式构建旅行短名单。
架构
记忆存储在 ~/tripadvisor/ 中。设置方法请参见 memory-template.md。
text
~/tripadvisor/
├── memory.md # 偏好和重复性约束
├── sessions/
│ └── YYYY-MM-DD.md # 搜索上下文和候选选项
├── api/
│ ├── location-cache.md # 查询 -> location_id 映射
│ └── request-log.md # 脱敏后的端点、参数、状态、时间戳
├── shortlists/
│ └── {city}-{topic}.md # 带理由的排名选项
└── archive/
快速参考
memory-template.md |
| 官方 API 工作流 | api-workflows.md |
| UI 导航手册 | web-navigation.md |
| 条款与合规边界 | compliance.md |
核心规则
1. 仅使用官方 Tripadvisor 接口
仅使用官方 Tripadvisor API 端点和正常浏览器导航。切勿抓取隐藏数据、绕过访问控制或自动执行被禁止的数据提取。
2. 开始时明确选择模式
每个任务选择一种模式:
- - API 模式:结构化查询和机器友好字段
- UI 模式:交互式浏览和短名单指导
- 混合模式:API 用于发现,UI 用于最终验证
3. 在深度查询前解析 location_id
对于 API 模式,首先将用户查询映射到有效的 location_id,然后获取详情/评论/照片端点。将成功的映射缓存到 ~/tripadvisor/api/location-cache.md 中。
4. 优先使用 URL 驱动导航而非脆弱的点击
在 UI 模式下,在进行复杂的点击链之前,尽可能依赖稳定的 Tripadvisor URL(城市垂直页面和实体详情 URL)。
5. 安全处理同意和反机器人状态
如果出现 Cookie 对话框或反机器人插页,请求用户确认,记录拦截器,然后继续使用 API 模式或直接 URL。不要尝试绕过技术。
6. 生成可决策的输出
始终返回带有明确权衡的排名短名单:
7. 保持存储最小化和透明化
仅在 ~/tripadvisor/ 下存储可重复使用的旅行偏好和选定选项。在会话中首次写入时确认,避免敏感个人数据,切勿在日志中存储密钥(始终脱敏 API 密钥)。
常见陷阱
- - 在解析 location_id 之前开始深度详情调用 -> API 响应变得嘈杂或不可用。
- 依赖脆弱的 DOM 选择器 -> 当 Tripadvisor UI 更改时流程中断。
- 混合 API 和 UI 数据而不带时间戳 -> 过时的比较和错误的推荐。
- 忽略 Cookie/同意覆盖层 -> 自动点击失败并产生假阴性。
- 仅将评分视为真相 -> 遗漏评论时效性和重复投诉模式。
- 在查询中发送不必要的用户数据 -> 可避免的隐私风险和合规问题。
外部端点
| 端点 | 发送的数据 | 目的 |
|---|
| https://api.tripadvisor.com/api/partner/2.0/location/search | 目的地查询和语言范围 | 解析 locationid |
| https://api.tripadvisor.com/api/partner/2.0/location/{id} 及相关位置子路径 |
locationid、语言和轻量级过滤器 | 获取地点详情、评论、照片、附近上下文 |
| https://www.tripadvisor.com/* | 标准浏览器导航信号和用户搜索词 | 检查公开页面并验证短名单候选 |
没有其他数据被发送到外部。
安全与隐私
离开您机器的数据:
- - 目的地名称、可选日期窗口和轻量级过滤器发送到 Tripadvisor API 或网页。
保留在本地数据:
- - 偏好、短名单决策和 ~/tripadvisor/ 中的请求日志。
此技能不会:
- - 访问 ~/tripadvisor/ 之外的文件
- 默认存储支付或护照数据
- 使用抓取绕过、CAPTCHA 规避或反机器人规避
信任
使用此技能,查询数据将被发送到 Tripadvisor 服务。
仅当您信任 Tripadvisor 处理您提供的搜索词和旅行上下文时才安装。
范围
此技能仅:
- - 操作 Tripadvisor 发现和比较工作流
- 使用官方 API 优先调用和政策安全的网页导航
- 生成带有透明权衡的排名短名单
此技能从不:
- - 声称保证预订结果
- 将不确定数据呈现为已核实事实
- 在没有明确用户指令的情况下执行购买或账户操作
相关技能
如果用户确认,使用 clawhub install
安装:
- - booking — 比较住宿选项和总成本明细
- travel — 管理更广泛的旅行规划工作流
- apple-maps — 在 macOS 上验证路线摩擦和区域可达性
- search-engine — 改进查询迭代和来源三角验证
- expenses — 在短名单决策后跟踪旅行支出
反馈
- - 如果有用:clawhub star tripadvisor
- 保持更新:clawhub sync