mechatracker

The problem#

Mechanics juggle YouTube tutorials, PDF service manuals, parts lookups, and tribal knowledge across half a dozen tools, with no persistent record per vehicle. Solo mechanics, shop teams, and small fleet operators all want the same thing: one place that tracks vehicles, repairs, parts, and history — and that works on a phone next to a lift.

What I built#

A three-surface stack sharing a single contract:

• Go API server — HTTP server with sqlc-generated type-safe queries over Postgres, golang-migrate for migrations, JWT verification middleware for Clerk, and a separate background-worker binary for push notifications and long-running jobs.
• Next.js web app — React + TypeScript, consuming an API client auto-generated from the server’s OpenAPI spec.
• React Native + Expo mobile — same generated client, Expo Push → APNs/FCM for notifications, EAS Build/Submit for release.
• Infrastructure — DigitalOcean Droplets, Managed Postgres, Spaces (S3-compatible storage), DOCR, all declared in Terraform and deployed via GitHub Actions.

Repair cards are the core concept — recipe-style profiles with parts, tools, torque specs, and step-by-step procedures. Users save vehicles, drill into system → repair type, log step completions with timestamped notes and photos, and the finished repair gets archived as an immutable snapshot per vehicle.

A decision I’d defend#

I auto-generated the TypeScript API client from the Go server’s OpenAPI spec via openapi-typescript, so both the web and mobile apps consume the same typed client. Hand-maintaining types across three languages is where product velocity goes to die; this way an API change propagates to every surface in a single commit, and the TypeScript compiler surfaces breakage before the apps even start.

// api/handlers/repairs.go — the Go handler and its OpenAPI tag drive codegen
// @summary Create a repair card from a template
// @id      createRepairFromTemplate
// @tags    repairs
// @accept  json
// @produce json
// @param   body body CreateRepairFromTemplateRequest true "request"
// @success 201  {object} RepairCard
// @router  /v1/repairs [post]
func (h *Handler) CreateRepairFromTemplate(w http.ResponseWriter, r *http.Request) {
    // ...
}

// web — same shape, auto-generated, used in a React Query hook
import type { paths } from '@/lib/api/schema'
type CreateReq = paths['/v1/repairs']['post']['requestBody']['content']['application/json']
type RepairCard = paths['/v1/repairs']['post']['responses']['201']['content']['application/json']

export const useCreateRepair = () => useMutation({ mutationFn: (body: CreateReq) => api.POST('/v1/repairs', { body }) })

Where it stands#

The backend, web shell, mobile scaffold, and infrastructure are all in place. Progress stalled during the Clerk auth integration — custom JWT verification in the Go middleware plus session handoff into Expo turned into a longer rabbit hole than I’d planned for. The project is paused rather than abandoned; picking it back up is on the list once the auth design is settled.

The repo is private.