garage-tracker

Garage-tracker is a full-stack web application for tracking vehicle maintenance.

API (Services)

All endpoints are mounted under /api/services and return JSON.

Records

Method	Path	Description
GET	/api/services	List all service records.
POST	/api/services	Create a service record.
GET	/api/services/:id	Get one record by id.
PUT	/api/services/:id	Replace a record by id.
DELETE	/api/services/:id	Delete a record by id.

Filters (query params on GET /api/services, combinable)

• ?vehicleId=<id> — only that vehicle's services
• ?serviceType=<type> — only that service type
• ?from=YYYY-MM-DD&to=YYYY-MM-DD — only services in that date range (either end optional)

Reports (computed views, grouped under summary/)

Method	Path	Description
GET	/api/services/summary/by-vehicle	Total spend and service count per vehicle.
GET	/api/services/summary/monthly	Total spend and service count per month.
GET	/api/services/summary/due-soon	Each vehicle's predicted next service by mileage (milesLeft, negative = overdue), most urgent first.

Use of AI (Data Generation)

The seed data for the two collections was created in two steps:

1. Mockaroo. I used Mockaroo to generate the raw records as JSON:

• Vehicles (data/vehicles-mockaroo.json, 300 rows): make (Car Make), model (Car Model), year, currentMileage, purchasePrice, and status (a custom list of Active / In Repair / Garaged / Sold). No id field was added, so MongoDB generates each _id.
• Services (data/services-mockaroo.json, 700 rows): date (formatted YYYY-MM-DD), serviceType (custom list), mileageAtService, cost (2 decimals), recommendedInterval, shopName, serviceRating (1–5), and notes. The services were generated without a vehicleId, since the link is added in the next step.

2. Claude. I then used Claude (Anthropic) to write a Node script (data/loadServices.js) that prepares a test database from those two files. I directed it to:

• Insert the 300 vehicles, then read back their MongoDB _ids.
• Give each vehicle a random 0–5 services, so some vehicles have no service history.
• Link each service to a vehicle by storing that vehicle's _id as the service's vehicleId (the foreign key). vehicleId is stored as a native MongoDB ObjectId (the same type as vehicles._id), so the two collections join directly without any type conversion.
• Clamp each service's mileageAtService so it never exceeds the vehicle's currentMileage.
• Stop once 700 services are assigned, giving exactly 1000 total documents across both collections.
• Add a unique nickname to each vehicle.

Representative prompt

I have two mockaroo json files, one for vehicles and one for services. write me a node script
that loads them into mongo. insert the vehicles first so they each get an _id, then give each
vehicle a random 0 to 5 services and link them by putting the vehicle's _id as the service's
vehicleId. also make sure a service's mileageAtService isn't higher than the vehicle's
currentMileage. keep going till 700 services are added so both collections add up to 1000 total,
and print how many services each car got.

This script is a development convenience for populating a test database; in normal use the database is filled through the application's frontend.