旅行住宿指南
trip-guide-pdf-lodging
by allensu0314
Research, plan, revise, and deliver lodging-anchored travel guides as HTML/PDF with verified route data, hotel selection, fallback hotel swaps, curated screenshots, and formal copy. Use when a trip includes hotels, inns, guesthouses, resorts, multi-night stays, or when the user asks to make or revise an itinerary after a hotel changes, sells out, or becomes too expensive.
安装
claude skill add --url https://github.com/openclaw/skills文档
Trip Guide PDF Lodging
Build the guide in HTML first. Export PDF only after route logic, lodging choice, and screenshots are stable.
Treat the hotel or lodging as the nightly anchor. Design each day around where the user sleeps, not around scenic spots alone.
Core rules
- Verify every hard number before writing it.
- Separate hard data from soft signal.
- Hard data: driving time, distance, tolls, hotel address, phone, opening hours, scenic latest entry.
- Soft signal: renovation quality, noise risk, dining feel, queue risk, convenience impressions.
- If lodging changes, recompute all dependent legs instead of text-replacing the hotel name.
- Keep screenshots sparse and purposeful.
- Use formal, compact copy unless the user explicitly wants casual tone.
Workflow
1) Lock the planning frame
Extract:
- dates / trip length
- departure city
- trip style: self-drive, hiking, family, solo, etc.
- preference weights: scenery, comfort, food, crowd avoidance, budget
- output target: quick answer or polished HTML/PDF
If the user already gave enough constraints, start researching immediately.
2) Choose the lodging strategy first
Before writing the day plan, determine:
- single hotel vs multi-hotel
- town-center convenience vs scenic proximity
- parking requirement
- breakfast requirement if it changes departure time
- fallback lodging in case of sellout or price spike
The day plan should follow the lodging choice, not the other way around.
3) Choose sources by job
Read references/source-selection.md when deciding what to trust.
Default split:
- maps / official scenic pages / structured listings for hard numbers
- review sites and social posts for soft signal
- if Chinese travel/review sites are involved and normal search/fetch is weak, use
cn-review-sites-cdp
4) Verify hard data before drafting
Check the numbers that decide feasibility:
- origin → lodging
- lodging → main scenic spot
- lodging → secondary scenic spot
- lodging → dinner / breakfast anchors when they matter in the final guide
- scenic opening hours / latest entry
- hotel phone / rating / address if shown
Re-run this step after every lodging change.
5) Design the itinerary from the lodging anchor
For each day, choose the nightly lodging first, then build:
- arrival window
- check-in / rest buffer
- meal window
- scenic entry window
- return buffer
- holiday congestion buffer
If the hotel moves from old town to new town, update the walking radius and dinner logic. Do not assume Day 1 still works unchanged.
6) Draft HTML first
Good structure:
- cover / conclusion box
- route overview
- day-by-day plan
- lodging section
- dining section
- risk notes / contingency
The top conclusion should tell the user:
- which lodging won
- why it won
- what fallback exists
- what tradeoff changed after the lodging decision
7) Use screenshots as evidence, not decoration
Keep only screenshots that support a decision:
- scenic proof image
- hotel listing / hotel note if it materially affects the recommendation
- restaurant listing snippet if it justifies a recommendation
Reject screenshots that are blank, QR-heavy, cluttered, or mostly irrelevant UI.
8) Run the QA gate before PDF export
Read references/qa-checklist.md and clear it.
Minimum gate:
- route numbers consistent with latest lodging choice
- time logic feasible
- screenshots clean
- tone formal enough
- filenames and variant names clear
9) Deliver and version clearly
Prefer scenario-specific filenames over destructive overwrites.
Examples:
*_final.html*_revision.html*_hotel-swap.html*_quanji.html
When sending local files through OpenClaw messaging, prefer relative MEDIA:./... paths instead of absolute MEDIA:/abs/path paths.
Revision logic
- Visual-only feedback: rework screenshots, typography, and tone.
- Hotel change: recompute every dependent leg and rewrite the daily structure affected by that hotel.
- Budget pressure: rerank hotels and surface tradeoffs explicitly.
- Feasibility issue: re-verify with maps / official sources and rebuild affected days.
Read these references when needed
references/source-selection.md— which source to trust for which data typereferences/qa-checklist.md— pre-export checklist for lodging-based guides
相关 Skills
技能工坊
by anthropics
覆盖 Skill 从创建到迭代优化全流程:起草能力、补测试提示、跑评测与基准方差分析,并持续改写内容和描述,提升效果与触发准确率。
✎ 技能工坊把技能从创建、迭代到评测串成闭环,方差分析加描述优化,特别适合把触发准确率打磨得更稳。
PPT处理
by anthropics
处理 .pptx 全流程:创建演示文稿、提取和解析幻灯片内容、批量修改现有文件,支持模板套用、合并拆分、备注评论与版式调整。
✎ 涉及PPTX的创建、解析、修改到合并拆分都能一站搞定,连备注、模板和评论也能处理,做演示文稿特别省心。
PDF处理
by anthropics
遇到 PDF 读写、文本表格提取、合并拆分、旋转加水印、表单填写或加解密时直接用它,也能提取图片、生成新 PDF,并把扫描件通过 OCR 变成可搜索文档。
✎ PDF杂活别再来回切工具了,文本表格提取、合并拆分到OCR识别一次搞定,连扫描件也能变可搜索。
相关 MCP 服务
文件系统
编辑精选by Anthropic
Filesystem 是 MCP 官方参考服务器,让 LLM 安全读写本地文件系统。
✎ 这个服务器解决了让 Claude 直接操作本地文件的痛点,比如自动整理文档或生成代码文件。适合需要自动化文件处理的开发者,但注意它只是参考实现,生产环境需自行加固安全。
by wonderwhy-er
Desktop Commander 是让 AI 直接执行终端命令、管理文件和进程的 MCP 服务器。
✎ 这工具解决了 AI 无法直接操作本地环境的痛点,适合需要自动化脚本调试或文件批量处理的开发者。它能让你用自然语言指挥终端,但权限控制需谨慎,毕竟让 AI 执行 rm -rf 可不是闹着玩的。
EdgarTools
编辑精选by dgunning
EdgarTools 是无需 API 密钥即可解析 SEC EDGAR 财报的开源 Python 库。
✎ 这个工具解决了金融数据获取的痛点——直接让 AI 读取结构化财报,比如让 Claude 分析苹果的 10-K 文件。适合量化分析师或金融开发者快速构建数据管道。但注意,它依赖 SEC 网站稳定性,高峰期可能延迟。