mcpField
C
Cycle Tracker

Cycle Tracker

Remote
cycle.arrakis.houseMaps & Travel1Updated 14w ago

About Cycle Tracker

Cycle Tracker is a Manchester bike-share dashboard for people tracking station availability and usage. It combines live station data, an interactive map, and historical snapshots.

Key features

  1. 01Live bike and station availability
  2. 02Station capacity and utilization
  3. 03Interactive station map
  4. 04Popular-station analytics
  5. 0590-day history

Use cases

  1. 01Check nearby bike and station availability
  2. 02Compare station capacity and usage
  3. 03Review popular stations
  4. 04Explore recent availability trends

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://cycle.arrakis.house/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

5

This tool takes no parameters.

Details

From the project README

Starling Bank Bikes Manchester

Starling Bank Bikes Manchester is a dashboard that monitors bike-sharing stations in Manchester in real time and reports analytics about them. The project includes a web frontend and a backend API.

Features

  • Live bike and station stats: View availability, utilization, and capacity for each station.
  • Interactive map: See station locations and bike distribution.
  • Popular-station analytics: Find the most-used and highest-capacity stations.
  • Modern UI: Use a responsive interface built with React, Tailwind CSS, and Radix UI.

Project structure

- bike-sharing-dashboard/   # Frontend (React, TypeScript, Vite)
  ├── src/
  │   ├── components/
  │   │   ├── dashboard/         # Dashboard features (PopularStationAnalytics, StatsCards, InteractiveMap)
  │   │   ├── home/              # Home page
  │   │   └── ui/                # UI primitives (Sheet, Tabs, Card, Table)
  │   ├── lib/                   # Utility functions
  │   ├── App.css, index.css     # Styles
  │   └── main.tsx               # App entry point
  ├── index.html                 # Root HTML
  ├── vite.config.ts             # Vite config
  └── tailwind.config.js         # Tailwind config

- worker/                   # Backend (Cloudflare Worker, TypeScript, Hono, D1)
  ├── src/
  │   ├── index.ts              # Hono app, CORS, static assets, cron entry
  │   ├── snapshot.ts           # GBFS feed -> D1 snapshot job (10-min cron)
  │   └── routes/               # API routes (bikes, stations, analytics)
  ├── migrations/               # D1 schema migrations
  └── wrangler.jsonc            # Worker + D1 + cron + custom-domain config

Get started

Prerequisites
  • Node.js and npm
  • A Cloudflare account with Workers and D1, for deployment
Develop locally

To run the backend (Worker plus local D1), enter the following commands:

cd worker
npm install
npm run migrate:local   # first time only
npx wrangler dev        # API on http://localhost:8787

To run the frontend, enter the following commands:

cd bike-sharing-dashboard
npm install
npm run dev             # UI on http://localhost:5173, calls the local worker
Deploy

The app runs at cycle.arrakis.house. When you push a change to main that touches worker/ or bike-sharing-dashboard/, the .github/workflows/deploy.yml workflow builds the frontend, applies the D1 migrations, and runs wrangler deploy. The Worker serves the static frontend and the /api/v1 API from the same origin. A 10-minute cron job snapshots the Manchester Beryl General Bikeshare Feed Specification (GBFS) feed into D1.

Technologies used

  • Frontend: React, TypeScript, Tailwind CSS, Vite, Radix UI
  • Backend: Cloudflare Workers, Hono, D1 (SQLite), TypeScript

Author

ThisisBankole

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.