Fleet Priority
A subscription SaaS that gives small rental-car operators one place to track every vehicle's condition, cost, and paperwork — built solo, end to end, with Claude Code.
STATUS: LIVE
The Brief
Small rental fleets run on spreadsheets, text threads, and a shoebox of paper mechanic receipts. Tag renewals slip. Insurance lapses go unnoticed until a claim. And there's rarely any record of what a car actually looked like before it left the lot versus when it came back.
Fleet Priority replaces that with a single dashboard per operator: every vehicle's tag, insurance, mileage, payoff status, maintenance history, and condition in one place, with real accounts, real subscriptions, and real accountability — not a shared spreadsheet anyone can silently edit.
It ships as a two-tier subscription product (Basic and Pro), with Stripe handling billing and a feature gate separating the tiers, so it's a real business, not just an internal tool.
What It Does
Fleet Dashboard
Search · filter · sortEvery vehicle as a card: status, tag expiration, tires, mechanic, and a payoff bar that decays automatically each month from a monthly-payment figure, so balances stay current without manual updates. Search by tag, VIN, make, model, or mechanic; filter by status; sort by mileage, year, or tag.
Vehicle Record
One car, full historyMaintenance history with attached receipts, a per-vehicle gas-cost trend chart plotting every individual fill-up on a fixed axis, insurance documents and policy tracking (Pro), and a preferred-mechanic contact — all locked to view-only until an explicit Edit is pressed, so nothing changes by accident.
Vehicle Inspections
Newest moduleA fixed six-point photo checklist — front, rear, each side, interior, odometer — captured straight from a phone camera. The distinguishing move: starting a new inspection automatically pulls up the same vehicle's last inspection photos side by side, so damage gets judged against the last known state, not memory. Each finding carries a MINOR or MAJOR severity tag and a note.
Fleet-Wide Reporting
Spend across every carSpend broken down by service type, month-over-month oil-change and gas-cost trends, and a running log of recent maintenance across the whole fleet — the numbers an owner actually checks before deciding whether a car is worth keeping.
Notifications & Account
Nothing slipsTag, insurance, and maintenance deadlines surface automatically — overdue items flagged in red, due-soon in amber — with optional email reminders sent on a schedule. Settings covers profile, a theme picker that can follow the OS live, password changes, and a list of every signed-in device with per-device sign-out.
Under the Hood
The interesting parts weren't the CRUD screens. A few real problems came up along the way — here's how each got run down.
A cascading delete that quietly failed
Deleting a vehicle sometimes left its maintenance records behind. The cause: the data layer read the JSON file fresh, mutated it, and wrote it back — twice, in two separate calls — so the second write could clobber the first before it landed. The fix wasn't a patch on the delete route; it was restructuring the whole store around one shared in-memory copy with serialized writes, so every mutation is immediately visible to the next one regardless of disk timing.
Sessions, not just tokens
Auth started as a bare JWT — fine for "is this user logged in," useless for "sign this one device out." Reworked so every login mints a server-tracked session record and the token carries a session id instead of just a user id. That's what makes per-device revocation possible, and what lets a password change safely kill every other session in one move without guessing which token belongs to which browser.
A live Stripe integration, debugged live
Checkout worked in testing, then failed in production with Managed Payments is not supported on API version 2024‑06‑20. The account had a newer Stripe feature enabled than the SDK's default API version supported. Fix was to stop relying on the SDK's default and pin the API version explicitly — the kind of bug that only shows up against a real account, not a sandbox.
A real box, not a platform
Deployed to a plain Ubuntu VM rather than a managed platform, since the JSON-file store and uploaded photos need a persistent disk: Node process managed by pm2, nginx as reverse proxy, DNS delegated through Route 53, TLS via Let's Encrypt, and a live Stripe webhook endpoint wired up for real subscription events.
Stack
| Backend | Node.js & Express, REST API under /api/* |
| Data | JSON-file store, in-memory cache with serialized writes — documented seam to swap in Postgres at scale |
| Auth | JWT in an httpOnly cookie, bcrypt password hashing, server-tracked sessions for device management |
| Payments | Stripe Checkout, subscriptions, webhooks, promotion codes |
| Frontend | Vanilla JS, HTML, CSS — no framework — Chart.js for trend charts |
| Resend API for password resets & reminders, console-log fallback for local dev | |
| Infra | AWS Lightsail (Ubuntu), nginx, pm2, Let's Encrypt, DNS on Route 53 |