I got fed up building Tesla dashboards manually out of picture-elements — and waiting for the community to build the card I wanted — so I vibecoded the hell out of it. Tesla Fleet Card is now at v1.0 and ready for testing.
One card shows your whole Tesla fleet, looking and behaving like the official Tesla app. No picture-elements, no overlays to position, no advanced programming: four fields per car (name, model, paint, entity prefix) and you’re done. A dropdown on the car’s name flips between cars.
What it does
- Home view — your car resting, like the app’s opening screen; plugging in swaps to the cable shot, with “time remaining to charge limit” up top
- Controls view — top-down car with tappable frunk/boot (two-tap confirm), lock/unlock, tyre pressures at the corners (psi or bar), charge port
- Climate view — the interior: tap seats and steering wheel for heaters (uses whichever heat levels your car offers, ventilated seats included), temperature, Vent, Defrost, plus Bioweapon/Camp/Pet modes and Cabin Overheat Protection where supported
- Charging panel — charge-limit slider with a click-stop at 80%, live kW / +kWh / amps / volts, amps stepper, Stop Charging / Unlock Port
- Both integrations, automatically — works with tesla_custom (HACS) and the official tesla_fleet integration, auto-detected per car; mixed fleets are fine
- Images optional — drawn artwork in your paint colour works out of the box; image packs for Model 3 (grey) and Model Y (red) are included and load automatically, or make a pack from screenshots of your own car (the README shows how)
Install
- HACS → three-dots menu → Custom repositories → paste the repo URL, category Dashboard, Add
- Download Tesla Fleet Card, hard-refresh your browser
- Add the card, fill in the four fields
type: custom:tesla-fleet-card
cars:
- name: My Tesla
model: Model 3
paint: grey
prefix: ""
Full disclosure: this is fully vibecoded — built conversationally with Claude (Anthropic) against screenshots and screen recordings of the real Tesla app until the two were hard to tell apart. I’ve tested a load of it, but I’m certain there’s plenty people will find wrong — issues and PRs very welcome, or fork away to your heart’s content 
If anyone wants to join the project properly, let me know.
