Atlas agent
全球高级定制化 AI 旅游攻略 Skill Pack — 9 skills, zero API key, Hermes/Cursor/Claude compatible
npx -y skills add wpfbcr/AtlasAgent --skill atlas-agentAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Plans global travel itineraries with multi-plan comparison, day-by-day schedules, multi-currency budgets, visa/diet/transport guidance, and PDF export. Use when the user asks for trip planning, vacation itinerary, travel guide, 行程规划, 旅游攻略, or holiday planning for any destination worldwide.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
6.0 KB, as published. Nobody here has run it
AtlasAgent — Global Travel Planning
Orchestrates worldwide trip planning: multi-plan comparison, daily schedules, budgets, visa/diet/transport decisions, PDF export.
Install
If not installed yet, follow docs/install.md or tell the user:
帮我安装 AtlasAgent：https://raw.githubusercontent.com/wpfbcr/AtlasAgent/main/docs/install.md
Verify: bash "$ATLAS_ROOT/scripts/atlas-agent doctor"
When to Use
- User provides origin, dates, party size — needs an itinerary
- User wants destination recommendations with plan comparison
- User has booked flights/hotels and needs a concrete schedule
- Multi-city or cross-timezone trips
Required Inputs
Ask for missing fields before planning:
| Required | Optional |
|---|---|
| Origin (city + country) | Budget (total or per person + currency) |
| Dates + duration | Style (food, nature, history, urban, adventure, family, romantic) |
| Party size + relationship | Mobility/diet/photo/slow-travel preferences |
| Destination (or ask agent to suggest) | Passport nationality, booked tickets/hotels |
| Output language (default: 中文 if user writes in Chinese) |
Paths (required)
Install the full repository — skills/ alone is not enough.
export ATLAS_ROOT="${ATLAS_ROOT:-$HOME/AtlasAgent}"
export ATLAS_SCRIPTS="$ATLAS_ROOT/scripts"
export ATLAS_REF="$ATLAS_ROOT/skills/atlas-agent/references"
Install: see docs/install.md — full repo required, not skills/ alone.
Workflow
Phase 0 — Profile
- Passport nationality → read
visa-entryskill - Dietary restrictions → read
dietary-globalskill - Booked tickets/hotels → treat as fixed anchors
Phase 1 — Data (parallel)
# Weather
python3 "$ATLAS_SCRIPTS/weather_client.py" forecast --city "<city>" --days <N>
# Currency
python3 "$ATLAS_SCRIPTS/currency_client.py" rates --base <currency>
python3 "$ATLAS_SCRIPTS/currency_client.py" convert <amount> --from <A> --to <B>
# Calendar
python3 "$ATLAS_SCRIPTS/holiday_check.py" --city "<city>" --from YYYY-MM-DD --days <N>
Geography (first available):
$MAPS_SCRIPTif set (any geocoder/POI CLI, e.g. OpenStreetMap client)- Web search +
$ATLAS_REF/destinations/<city>.md - Never invent hotel/restaurant names
Flights/hotels (optional L1): trvl MCP or web search — see mcp/recommended-servers.md
Reviews (optional L1): local-intel skill
Phase 1.5 — Disambiguation
Do not guess ambiguous names. Present options and ask the user to choose:
- Location, transport, duration, ticket/booking needs
- Local + English names
Examples: Paris (France vs Texas), Springfield (30+ US cities).
Phase 2 — Scheduling
- Use
holiday_check.py— weekdays, holidays, museum closure hints - Book major sights on weekdays when possible
- Arrival day = half day; allow jet-lag buffer after eastbound flights
- Multi-city: minimize backtracking (use distance/routing tools if available)
- Respect booked transport times as hard constraints
Phase 3 — Itinerary
Each plan includes:
- Round-trip transport (route, time, price in local + user currency)
- Weather summary (Open-Meteo)
- Visa/entry summary (from visa-entry)
- One-line trip theme + destination timezone
Daily table:
| Schedule | Stay | |
|---|---|---|
| D1 date (weekday) | AM → PM → evening (hours) | area/hotel |
| … | … | … |
Required sections: must-eat list, 2–3 stays, budget table (budget-optimizer), transport decision (transport-global), document checklist (travel-documents), cautions.
Meals: every meal = restaurant + dishes with dietary tags (🟢🟡🔴⚠️) + price per person.
Pace: only one "intense" day; others relaxed. Mark activity hours.
Comparison rubric: references/comparison-rubric.md
Phase 4 — Compare & Recommend
Multi-plan table (days, budget, travel time, romance/food/photo/culture/family stars, effort, visa complexity). End with 1–2 sentence ranking.
Phase 5 — Export
python3 "$ATLAS_SCRIPTS/md2pdf.py" itinerary.md
Capability Tiers
| Tier | Sources | Label in output |
|---|---|---|
| L0 | Scripts + $ATLAS_REF/destinations/ | Default |
| L1 | + trvl MCP, social research | Cite source + query time |
| L2 | + Amadeus, calendar/docs export | Cite API + query time |
If data is estimated, say: 「基于 reference 估算，出行前请验证」.
Companion Skills
| Skill | Purpose |
|---|---|
weather | Forecasts |
currency | FX conversion |
visa-entry | Visa/entry |
dietary-global | Diet restrictions |
transport-global | Transport choices |
local-intel | UGC research |
travel-documents | Checklists |
budget-optimizer | Budget format |
Pitfalls
- Rate-limit POI APIs (>2s between calls); fall back to references
- Visa/policy: always note query date; re-check before departure
- FX: note rate date (Frankfurter/ECB)
- Southern hemisphere seasons invert vs northern
- Tipping varies by country → budget-optimizer
- Never recommend conflicting transport if user already booked
Verify
python3 "$ATLAS_SCRIPTS/weather_client.py" forecast --city Paris --days 3
python3 "$ATLAS_SCRIPTS/currency_client.py" convert 100 --from EUR --to CNY
python3 "$ATLAS_SCRIPTS/holiday_check.py" --city Paris --from 2026-07-14 --days 5