Routekit

Plan gravel rides with elevation you can feel

Théo MartinLaunched 3 weeks agoWeb appWorking · checked 3 hours ago
0:42

About

Draw a route, and Routekit snaps it to gravel and forest roads, shows the climbs as colour on the map and estimates time based on your own last ten rides.

Open source and self-hostable. Bring your own Mapbox key.

How it was made

Shared by the maker. Numbers are their own estimate.

Build time
3 weeks
AI spend
$28
Prompts
~120
Written by AI
70%

Agents

  • Cursor

    50% of the work

    Map interactions and GPX export

  • Claude

    50% of the work

    Elevation smoothing and time model

MCP servers

No MCP servers. Built with the agents alone.

Build recipe

The code is on GitHub. If you’d rather build it yourself than fork it, paste this into your own agent.

Written by Claude

From the project’s code and history · 350 words

Build a web app called Routekit: plan gravel rides with elevation you can feel.

What it does

Draw a route, and Routekit snaps it to gravel and forest roads, shows the climbs as colour on the map and estimates time based on your own last ten rides.

Stack

Use Next.js, Mapbox, Supabase. The original was built with Cursor and Claude in 3 weeks, for about $28 in AI usage.

How it fits together

  • A server-rendered web app with a small API layer
  • A database for user data, with row level security on every table
  • Sign-in so each person only sees their own data

Build it in this order

  1. Set up the project, the database and sign-in, and deploy an empty page
  2. Build the main screen with realistic sample data
  3. Connect it to real data and add the one action the app exists for
  4. Add the empty states and error messages a new user will hit
  5. Write tests for anything that touches money, dates or other people's data

A prompt that worked: Estimate ride time from my own history

Here are my last 10 rides as GPX summaries (distance, elevation gain, surface mix, moving time). Fit a simple model I can explain to a user in one sentence, that predicts moving time for a new route. Show the model, the error on each ride, and the sentence.

Watch out for

  • Don't build settings, teams or billing before the main screen works
  • Keep the first screen useful with sample data, so the app makes sense before anyone has set it up

Done when

  • A new user can sign in and get value in under two minutes
  • Everything works on a phone
  • Tests pass for the core logic

Before writing code, ask me up to five questions about who this outdoors web app is for and what the first version must do. Then work in small steps: after each one, tell me how to try it and wait for my feedback.

Discussion

4 comments · 1 built their own

  • Noah FischerBuilt one1 day ago

    Rebuilt it from the recipe and had it running in under an hour. Would you take a PR that adds a Docker set-up to the original?

  • Amara Okafor2 days ago

    Copied the main prompt and adapted it for a project at work. Saved me an afternoon.

    • Théo MartinMaker2 hours ago

      Thanks! Good question, I'll write it up in the build log this week.

  • Kenji Mori1 day ago

    What did the agent struggle with most? Always curious where people had to step in.