A parking-aware co-pilot. An MCP server that plans a short nearby trip on a given day — somewhere not too far from an anchor point, weather-appropriate, with suitable parking — so you never have to fiddle with Google Maps while driving.
You finish a call at 14:20. You've got two hours before the light goes. You want to park up somewhere with a view, maybe a short walk, ideally no rain, and somewhere the vehicle actually fits. So you open Google Maps, zoom around, pinch a coastline, try to remember which car park had the height barrier, check the weather in a second tab, work out sunset in your head. All while half-driving, half-typing at a red light. It's a rubbish way to decide — and it's how most of us actually do it.
Nomad Trip Planner takes the question — "where should I go for a couple of hours from here?" — and gives a direct answer with a parkup recommendation, the weather window, sunset time, and a Maps deeplink you can tap once. No scrolling.
"I've got 90 minutes between calls. Somewhere scenic. Pick a spot for me."
Calls plan_day_trip with your home/work anchor. Three candidates are ranked by weather window + drive time + vibe match.
One short paragraph, TTS-ready. Destination, sunset, weather, parkup notes. No screen.
A single tap opens the directions in Google Maps. You drive. The agent's already thinking about dinner.
Ranked candidates within your drive-time budget, filtered by vibe — beach, forest, mountain, quiet, scenic, or new-to-me.
Open-Meteo hourly forecast per candidate, then the server picks the window with the driest, warmest, least-windy conditions.
Computed locally with a NOAA approximation — no API calls, works offline, correct to the minute.
Height limits, surface type, overnight rules, facilities, and an honest one-line "best for" description.
Haversine distance times a road-factor, at an assumed 60 km/h. Good for ranking — for final ETA, Google Maps takes over.
Parkups with the scenic or new-to-me tag bubble up first when you ask for the longer route.
Desk research pass (no signups or scraping): the strongest Brit Stops / Park4Night-style products win by reducing uncertainty at the exact moment a van-lifer is deciding where to stop. Nomad should borrow the trust cues, not the clutter.
Members stay at pubs, farms, vineyards and rural businesses. Key cues: permission, arrival rules, purchase etiquette, facilities, dog friendliness, and big-rig suitability.
Photos, recent comments, height barriers, police/no-overnight reports, price, surface, noise and safety are often more useful than a generic star rating.
Travellers need water, waste, laundry, showers, fuel, repair help and offline maps. Nomad can surface “what you need next” alongside the pretty stop.
These lean into landscapes, photos, amenities and confidence. Useful inspiration: stronger nature-first visuals, filters for fire pits, walks, views and quiet hours.
Nomad Trip Planner speaks MCP over stdio. Weather uses Open-Meteo — free, no key — and parkups come from a local fixture today, with iOverlander / Park4Night / Google Places provider stubs ready to wire up. Register it with your MCP client:
{ "mcpServers": { "nomad-trip-planner": { "command": "node", "args": ["/path/to/mcp-servers/nomad-trip-planner/dist/index.js"], "env": { "NOMAD_WEATHER_PROVIDER": "open-meteo", "NOMAD_PARKUP_PROVIDER": "fixture", "NOMAD_DEFAULT_ANCHOR": "51.4545,-2.5879" } } } }
Then ask your agent: "Plan my Friday — 60 minutes max, somewhere scenic, avoid rain." It fires plan_day_trip, hands you three ranked options with Maps deeplinks, and a short TTS-ready brief.
Designed to be used parked, or with voice — never while typing. The agent outputs one paragraph short enough to read aloud in fifteen seconds, and one link short enough to tap once. The dashboard includes a Read brief aloud button that uses the browser's speech synthesis so you can hear your plan without looking at the screen.