# Tesla MCP Server

> What a Tesla MCP server is, the 40 tools mytesla.io exposes, which AI clients it works with, and how the Fleet API connection actually works.

A Tesla MCP server lets any MCP-compatible AI assistant — Claude, ChatGPT,
Cursor, VS Code, or a custom agent — read your car's status and send it
commands, through Tesla's official Fleet API. mytesla.io is one: this page
explains what that means and how it actually works.

## What MCP is, briefly

[Model Context Protocol](https://modelcontextprotocol.io) is the open
standard AI assistants use to call external tools during a conversation.
Instead of every AI vendor building its own one-off Tesla integration, an
MCP server exposes a fixed set of tools once, and any MCP-aware client —
Claude, ChatGPT, Cursor, VS Code, or your own agent — can call them the same
way. You don't install a Tesla-specific plugin per assistant; you point your
assistant at one URL.

## How mytesla.io's server works

1. You connect your Tesla to mytesla.io through Tesla's own OAuth consent
   flow — the same kind of authorization Tesla's own app uses, via the
   official **Fleet API** and **Vehicle Command Protocol**.
2. mytesla.io runs as a hosted MCP server at `https://mcp.mytesla.io/mcp`,
   speaking Streamable HTTP MCP with OAuth 2.1 authorization.
3. Your AI client connects to that URL, authenticates as you, and gets a list
   of available tools.
4. When you ask something in plain English, the AI decides which tool(s) to
   call and with what arguments, calls them, and reports back.

No location, drive, or charge history is stored on mytesla.io's side — data
is fetched live per request and handed back. See the
[privacy policy](/privacy) for the full detail.

## The 40 tools

- **Reads:** `get_vehicles`, `get_vehicle_status`, `get_credit_balance`,
  `wake_vehicle`
- **Climate:** `start_climate`, `stop_climate`, `set_cabin_temperature`,
  `max_defrost`, `set_seat_heater`, `set_steering_wheel_heater`,
  `set_seat_cooler`, `set_climate_keeper_mode`,
  `set_cabin_overheat_protection`
- **Charging:** `start_charging`, `stop_charging`, `set_charge_limit`,
  `set_scheduled_charging`, `set_scheduled_departure`, `set_charging_amps`,
  `open_charge_port`, `close_charge_port`
- **Access & cargo:** `lock_vehicle`, `unlock_vehicle`, `actuate_trunk`,
  `control_windows`, `sun_roof_control`
- **Alerts:** `honk_horn`, `flash_lights`
- **Navigation & smart-home:** `share_destination`, `trigger_homelink`
- **Security:** `set_sentry_mode`, `set_valet_mode`, `reset_valet_pin`,
  `set_speed_limit`, `speed_limit_activate`, `speed_limit_deactivate`,
  `speed_limit_clear_pin`
- **Software:** `schedule_software_update`, `cancel_software_update`
- **Feedback:** `report_bug` (always free)

Every tool maps directly to a real Fleet API or Vehicle Command Protocol
call — there's no "do anything" escape hatch, so the AI is bounded to exactly
this list.

## Supported clients

Any client that speaks Streamable HTTP MCP with OAuth 2.1 works. In practice,
that's:

- **Claude** (Claude.ai web, Claude Desktop, Claude Code) — see
  [Control your Tesla with Claude](/control-tesla-with-claude)
- **ChatGPT** (Pro/Team/Enterprise/Edu with Developer Mode) — see
  [Control your Tesla with ChatGPT](/control-tesla-with-chatgpt)
- **Cursor** and **VS Code** — one-click install links are on the
  [home page](/#install)
- Any other MCP-aware client, pointed at `https://mcp.mytesla.io/mcp`

The server publishes standard OAuth discovery metadata at
`https://mcp.mytesla.io/.well-known/oauth-protected-resource` and
`.../.well-known/oauth-authorization-server`, and an MCP server card at
`https://mytesla.io/.well-known/mcp/server-card.json`, so clients that support
autodiscovery can find and configure it automatically.

## Pricing

Access is a flat monthly subscription, not per-call metering:

- **Driver** — $4.99/month, 500 credits
- **Driver Max** — $9.99/month, 1,500 credits

Most commands cost 1 credit; waking a sleeping car costs 10 (Tesla itself
rate-limits and prices wake calls more heavily). `get_vehicles` and
`report_bug` are always free. Full detail in the
[billing policy](/billing-policy).

## Which Teslas are supported

Any Model S, 3, X, or Y that supports Tesla's Fleet API and Vehicle Command
Protocol — in practice, essentially every Model S/3/X/Y from 2018 onward.
Some tools are trim-specific (`sun_roof_control` on older panoramic-roof
Model S, `set_seat_cooler` on ventilated-seat trims) — the AI tells you when
a command doesn't apply to your car rather than failing silently.

## FAQ

**Is this an official Tesla product?**
No. mytesla.io is an independent product operated by Embay, LLC, built on
Tesla's *official* Fleet API — but not built, endorsed, or operated by Tesla,
Inc.

**Does the AI ever get more access than I explicitly grant?**
No. The OAuth scope you approve on Tesla's consent screen sets the ceiling;
mytesla.io can't request anything beyond it, and it only exposes the 40 tools
listed above regardless of what the underlying Fleet API supports.

**Can I use my own MCP client instead of Claude or ChatGPT?**
Yes — anything that supports Streamable HTTP MCP with OAuth 2.1 can connect
to `https://mcp.mytesla.io/mcp`. The [full setup guide](/docs) has more on
authentication.

## Next steps

- [Control your Tesla with Claude](/control-tesla-with-claude)
- [Control your Tesla with ChatGPT](/control-tesla-with-chatgpt)
- [25+ things you can do with Tesla + AI](/tesla-ai-use-cases)
- [Full setup guide](/docs) — every command, credit cost, and troubleshooting
  step
- [Get started at mytesla.io →](https://mcp.mytesla.io/signup?src=guide-mcp&plan=driver)