travel
Skill0 starsChanged 2 months agoMITView on GitHub
Turn a traveler's raw materials (bookings, notes, screenshots, links, chat messages) into a personal, shareable travel website ("roadbook") using the dependency-free template in this repo. Use when asked to build, import, update, check, or publish a trip roadbook / travel site / itinerary page from travel materials.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- mariagorskikh/open-instincttheir own version, first seen Oct 14, 2026
- alexjmiller5/agent-config-publictheir own version, first seen Oct 2, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- mariagorskikh/open-instinct★ 428same name
- alexjmiller5/agent-config-public★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/travel-roadbook.mdcCursor rule
- design-with-aiSkill
- travel-managementSkill
- travel-product-studioSkill
- travelSkill
- travel-template-studioSkill
- travel-verificationSkill
- design-with-aiSkill
- travel-managementSkill
- travel-product-studioSkill
- travelSkill
- travel-template-studioSkill
- travel-verificationSkill
- design-with-aiSkill
- travel-managementSkill
- travel-product-studioSkill
- travel-template-studioSkill
- travel-verificationSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Run these in order. Each step has a command and an exit condition.
- Very specific: line 40
Facts come from materials or the user: No invented departure minutes, prices, addresses, opening hours, or booking status.
- Distinctive: line 51
Read text first (PDF text, emails, chat exports, notes).
- Very specific: line 53
Keep the local-language name of every place (localName) and an address the traveler can show to a driver.
- Very specific: line 71
The traveler's own photos first: Ask for them. Resize before committing (cover ≤ 200 KB, thumbnails ≤ 60 KB.
- Distinctive: line 73
Never let an image carry a fact.
- Distinctive: line 80
Read only what the current step needs.
- Very specific: line 95
If the user asks for one specific change ("move the museum to Tuesday", "the hotel in Berlin is now booked"), edit the data, run npm run …
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
travel.skill — compatibility entry
This command-oriented entry remains supported. The complete five-skill system is routed by AGENTS.md: planning, product development, design, template/module extension and travel verification. For new templates or features use travel-template-studio; for complex travel use travel-verification.
From materials to a roadbook
Promise: the traveler drops their materials in, and leaves with a site they can open on their phone and share. You do the reading, structuring, gap-finding, previewing and release checks. You never invent facts.
Success for the first version: a stranger can finish their first roadbook with you and this repo, without asking the author anything.
The pipeline
Run these in order. Each step has a command and an exit condition. Do not skip the gap step.
| Step | What you do | Command | Leave when |
| 0 Start | Copy an example into trip/ if there is no trip/travel-data.json yet. Pick the example closest to the trip (hiking / rail / family+car). | npm run new -- <example> | trip/travel-data.json exists and npm run dev shows it |
| 1 Import | Read everything in input/ (or wherever the user put materials). Extract facts into trip/travel-data.json. Keep the source of every fact in sources and sourceRefs. | — | Every fact you wrote can be traced to a file, a message, or "user said" |
| 2 Gaps | Run the validator and the gap report. Turn gaps into short questions. Ask them in one batch. | npm run validate · npm run gaps | Validator has 0 errors; each remaining gap is either answered or explicitly left "unknown" |
| 3 Generate | Fill the data with the answers. Choose trip.theme, trip.locale, trip.map.provider. Set trip.demo to false only when it is a real trip. | npm run validate | 0 errors; the user's answers are in the data, not in your notes |
| 4 Preview | Start the dev server and walk the five pages yourself: Home, Map, Days, Transport, Checklist. Open a map, mark a ticket, add a checklist item, reload. | npm run dev → http://localhost:4173/ | You have actually looked at each page; issues are fixed in data or reported |
| 5 Publish | Build, audit, then hand over dist/ with the publishing options. Do not upload anywhere without explicit permission. | npm run build · npm run check | Audit passes; the user knows what is public and how to update it |
Report at the end of every session in this shape:
Done: <what the user can open now, with the URL or folder>
Not done: <gaps still open, checks not run>
Entry: <local preview / dist folder / live URL, and which data version>
Next: <one concrete action for the user>Rules that do not bend
- Facts come from materials or the user. No invented departure minutes, prices, addresses, opening hours, or booking status. Unknown is
null,"unknown", or a gap — never a plausible guess. - Status vocabulary is closed:
unknown·needs-confirmation·booked·demo. A trip description is not a booking. Do not writebookedwithout a confirmation in the materials. - Times carry offsets and IANA zones. Flights use
2031-10-23T19:30:00-04:00plusorigin.timeZone/destination.timeZone. Durations are computed from absolute instants; the template already does this. Never subtract clock faces. - Public data is public.
travel-data.jsonindist/can be downloaded by anyone with the link. Passport numbers, booking references, PNRs, ticket numbers, payment details, private phone numbers and private addresses never go in. The validator rejects the obvious cases; you check the free text. - Do not touch
input/. Original materials are read-only. Do not copy them intotrip/ordist/. - The look is a choice, not a default. Three themes ship (
field-notes,timetable,tide). None of them is "the" look; pick by mood and destination, or add a theme. Do not bake one country's symbols or transit habits into another trip. - Evidence has levels. "Validator passed" is a structure check. "I opened the page and clicked" is runtime observation. "The user opened it on their phone" is adoption. Say which one you have.
- Authority stops at the repo. Building and previewing is yours. Creating accounts, pushing to a remote, deploying, buying anything, or sending messages needs the user's explicit go-ahead each time.
Reading materials well
- Read text first (PDF text, emails, chat exports, notes). Look at screenshots only when text is missing or contradictory.
- Distinguish booked (confirmation number exists — record the fact, not the number), planned (the user intends it), and suggested (a blog, a friend). Only the first two go into
days; suggestions can becomegroundTransportnotes or checklist items, marked as suggestions. - Keep the local-language name of every place (
localName) and an address the traveler can show to a driver. Add coordinates when you can verify them from a public source; otherwise usemapQuery. - Conflicts (two different hotel names for one night, a flight that lands after the train leaves) are gaps, not decisions you make.
Data shape
trip/travel-data.json — one file, validated by scripts/lib/validate.mjs. Full field reference: docs/DATA.md. The parts you will touch most:
trip— id, title, dates,timeZone,locale(enorzh-CNship),theme,people,map.provider(google|osm|none),demo.flightJourneys[]— one card per journey,segments[]per leg.places[]— everything a reader might want on a map.days[].events[].placeIdspoint here.accommodations[]— one per stay, linked fromdays[].accommodationId.days[]— one entry per calendar day of the trip,events[]in order,timeLabelfree text ("09:00"or"Morning").tickets[],groundTransport.tabs[],rental(only when there is one),checklist[],sources[].
Images (optional, do them last)
The roadbook works with no images. Add them only after the facts are right, in this order: cover (trip.cover.image, wide), one square thumbnail per day (days[].cover), a map-page photo (routeOverview.image), a paper tile (trip.textures.paper). Files live in trip/assets/ and are referenced as assets/...; only referenced files are published.
- The traveler's own photos first. Ask for them. Resize before committing (cover ≤ 200 KB, thumbnails ≤ 60 KB;
scripts/optimize-images.pyif Pillow is available, otherwise any tool). Do not publish photos of other people without asking. - No photos? If an image-generation tool is available in your environment, offer it and follow docs/VISUALS.md: realistic photograph of one subject per day, no text, no faces, no logos; record the prompt and model in
trip/ASSETS.md. If no tool is available, say so and ship without images; do not download stock photos of unknown license. - Never let an image carry a fact. Times, prices, statuses stay in the data.
- For a share card, set
trip.siteUrlto the final address soog:imageis absolute.
Budget: the template itself is under 100 KB; a cover adds about 100 KB before first paint; everything else loads lazily.
When to read the references
Read only what the current step needs.
- Imported materials and now structuring facts, time zones, currencies, what stays private → data-and-countries.md
- Building or changing a page, or judging whether a module is "done" → module-contracts.md
- Choosing or tuning the look, images, fonts, motion, mobile vs desktop → visual-and-motion.md
- Reviewing before handover, scoring, regression list, evidence levels → evaluation.md
- Publishing, caching, regional reachability, rollback → release-and-access.md
- Large trip or several people/agents working in parallel → orchestration.md
- Rental car / vehicle in the trip, or building a coverage table → coverage-and-rental.md
- Testing this skill's own judgement → scenarios.md
Copy-ready prompts for the user side of each step are in prompts/.
Small changes
If the user asks for one specific change ("move the museum to Tuesday", "the hotel in Berlin is now booked"), edit the data, run npm run validate, reload the preview, and confirm the visible result. Do not re-run the whole pipeline or re-ask settled questions.
What this skill does not do
- It does not book, pay, or contact vendors.
- It does not verify that a timetable is current; it records where a fact came from and asks the user to check time-sensitive items.
- It does not sync checklists between travelers. Reader checks live in each browser's local storage; if the user needs shared state, say so and stop — that is a different product.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

