Tesla MCP Server: https://mcp.myteslamate.com/mcp

This repository provides a Model Context Protocol server that wraps both the Tesla Fleet API and the TeslaMate API. It includes vehicle endpoints, vehicle commands, energy endpoints (Powerwall/Solar), and charging endpoints. The implementation is modular and fully featured, exposing 98 MCP tools for the Tesla Fleet API and 9 MCP tools for the TeslaMate API.

✨ What You Can Do

This isn't just remote control — it's a conversational interface to your entire Tesla ecosystem. Express your use cases directly to your AI — describe what you want in plain language and let it figure out which tools to call, in what order, and with what parameters. No app to open, no menu to navigate.

⚡ Energy Optimization

"My Powerwall is at 40%. Should I charge my car now or wait for solar production this afternoon?"

"Analyze my last 30 days of energy data — am I losing money by exporting to the grid while paying to charge at night?"

"Set the Powerwall backup reserve to 20% and schedule the car to charge from 2am to 6am at 16A."

The AI can cross-reference your solar production, Powerwall state, grid rates, and vehicle battery level to give you a recommendation — and act on it immediately.

🚗 Smart Vehicle Management

"I have a road trip to Lyon next Thursday. Based on my usual efficiency, will I need to stop for charging? Find the best Supercharger on the route."

"Compare my energy consumption this month vs last month. What changed?"

"Pre-condition the car at 7:45am every weekday and set the charge limit back to 80% — it's currently at 100% from the trip."

📊 Data & Insights (with TeslaMate API)

"What's my average km/kWh over the last 3 months? Is it getting worse?"

"Show me my most expensive charging sessions this year and where they happened."

"How many hours per week is the car sitting with climate running? What's the energy cost?"

🔐 Security & Peace of Mind

"Is my car locked and is Sentry mode on? I forgot to check before my flight."

"My car is parked downtown. Turn on Sentry mode and send me the current location."

🎨 Generative UI — Custom dashboards on demand

Beyond pre-built widgets, the server exposes a Generative UI capability so the AI can build interactive Prefab UI components on the fly to answer with charts, tables and metric cards — not just text.

"Show me a bar chart of my charging cost per location for last month."

"Build a dashboard for this car: current SoC, last 7 drives, and battery health degradation."

"Compare my efficiency this winter vs last summer with two side-by-side metric cards."

The AI uses generative_generate_prefab_ui to author Python code in a sandbox (Pyodide + Deno on the server); the resulting widget is rendered directly in your chat through a Skybridge-mime iframe — no template authoring on your side, no extra round-trip. The companion generative_search_prefab_components lookup tool keeps the LLM honest about component signatures.

It works equally well for Tesla Fleet API and TeslaMate users — the same generative tool is gated by either subscription.

Skills — pre-framed analytical workflows

The server also ships MCP Skills: markdown workflows the LLM reads on demand to chain tools into a coherent answer. The first one available is drive-efficiency-coach:

"Run the drive efficiency coach for the last 30 days and show it as a report."

The skill walks the AI through fetching your TeslaMate drives, computing Wh/km percentiles, identifying outliers, cross-referencing speed/temperature/trip-length, and rendering a Generative UI report (metric cards + outliers table + 1–3 actionable coaching insights anchored in your own numbers).

🤝 Contribute a skill

Skills are the easiest way to extend this server without touching Python. Pull requests welcome — drop a new directory under tesla_mcp/skills/ and SkillsDirectoryProvider picks it up automatically.

A good skill is:

1. Specific — answers one concrete question well rather than trying to do everything (e.g. "summarize my charging cost by location for a period" beats "manage my charging").
2. Tool-aware — names the exact MCP tools to call (teslamate_get_car_drives, energy_live_status, …) and the right arguments.
3. Numerically anchored — the coaching/insights step quotes numbers derived from the user's own data, not generic platitudes.
4. Visual when it helps — wraps up with a generative_generate_prefab_ui render for analytics/reports; plain text for one-shot questions.
5. Honest about pitfalls — calls out unit ambiguities, data gaps, edge cases the LLM should handle.

Minimal layout:

tesla_mcp/skills/your-skill-name/
└── SKILL.md

SKILL.md starts with YAML frontmatter (name, description) followed by the workflow. See drive-efficiency-coach/SKILL.md as a reference.

Ideas waiting for an author:

• charging-cost-breakdown — group last month's charges by location, surface cost-per-100km
• weekly-tesla-digest — drives + charges + battery snapshot + software updates as a Monday-morning email-style report
• range-anxiety-checker — given a destination, cross-reference SoC, weather, elevation and recent efficiency to call out a stop or reassure
• home-charging-window-finder — using TOU settings + recent charge sessions, find the cheapest 4-hour window to plug in tonight

Available capabilities (98 tools for Tesla Fleet API + 9 for TeslaMate API + 2 Generative UI tools + skills):

• Vehicle control: lock/unlock, climate, trunk, windows, lights, horn
• Charging: start/stop, set limits, schedule, charging amps, charge history
• Powerwall & Solar: live status, energy history, backup reserve, storm mode, grid import/export, time-of-use settings
• Navigation: send destinations, find nearby Superchargers
• Security: Sentry mode, valet mode, speed limits, PIN to drive
• Data & analytics: vehicle data, trip history, drive statistics (TeslaMate API)
• Generative UI: on-demand Prefab dashboards (charts, metrics, tables) authored by the AI in a sandboxed Pyodide runtime
• Skills: pre-framed analytical workflows (e.g. drive-efficiency-coach) the AI reads on demand to chain tools into a coherent answer

🔐 One-Click SSO — Connect with Your Tesla Account

The hosted server supports OAuth 2.0 Single Sign-On via Tesla's official authorization server. Compatible clients (ChatGPT, Claude.ai, VS Code…) trigger an automatic sign-in flow — no token copy-paste required.

MCP server URL: https://mcp.myteslamate.com/mcp

The server proxies the OAuth flow to auth.tesla.com, then exchanges your Tesla token for a MyTeslaMate token transparently. Your Tesla credentials never touch this server.

🤖 Setup on ChatGPT (Apps / MCP Connector)

ChatGPT supports native MCP connectors via Settings → Apps.

1. In chat.openai.com, open Settings → Apps.
2. Click Create app.
3. Enter the MCP server URL:
   Code

   Copy

   https://mcp.myteslamate.com/mcp
   

4. ChatGPT will auto-discover the OAuth metadata and prompt you to Connect.
5. Sign in with your Tesla account — the authorization is handled automatically.
6. The connector appears in your Enabled apps list and is available in all your chats.

🧠 Setup on Claude.ai (MCP Connector)

1. Open claude.ai → Settings → Connectors.
2. Click Add custom connector.
3. Enter a name (e.g. Tesla) and the MCP server URL:
   Code

   Copy

   https://mcp.myteslamate.com/mcp
   

4. Click Add — Claude will auto-discover the OAuth metadata and show a Connect button.
5. Click Connect — you'll be redirected to Tesla's login page.
6. After authorization, Claude will have access to all Tesla MCP tools.

Note: OAuth-based SSO requires a MyTeslaMate account. The server validates your Tesla token and links it to your MTM subscription automatically.

🤖 Claude Code Skills

Ready-to-use Claude Code skills for this MCP server are available in myteslamate/tesla-skills.

Install the /tesla slash command in one step:

Then in Claude Code:

/tesla what is my battery level?
/tesla lock my car
/tesla how much did I charge last month?

🚀 Quick Start - Use Directly in VS Code, Claude Desktop, etc.

No installation required! Use the hosted MCP server directly. Two authentication options:

Option A — SSO (Recommended for Claude.ai & ChatGPT)

Simply provide the server URL and let the client handle the OAuth flow automatically (see sections above).

Option B — Manual Token (VS Code, Cursor, Claude Desktop)

For clients that don't support OAuth, use your MyTeslaMate API token directly in the Authorization header.

For VS Code / Cursor (with MCP Extension) and Other MCP Clients

Add both endpoints to your .vscode/mcp.json configuration to choose your preferred integration:

{
  "servers": {
    "tesla": {
      "type": "http",
      "url": "https://mcp.myteslamate.com/mcp"
    }
  }
}

• Use tesla_fleet_api for direct access to Tesla's official Fleet API.
• Use teslamate for integration via TeslaMate API.

For Claude Desktop

Add endpoints in the MCP Connector configuration with the same URLs and headers as above.:

claude mcp add --transport http tesla-api "https://mcp.myteslamate.com/mcp"

Select the server according to your authentication method and API preference.