> ## Documentation Index
> Fetch the complete documentation index at: https://docs.volteras.com/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP server

> Ask questions about your connected fleet from any MCP-compatible AI tool.

The Volteras MCP server lets AI assistants read your fleet data directly. Connect it once and you can ask things like "which vans are below 20% charge?" or "how many vehicles need reconnecting?" in plain language, without writing API calls or exporting spreadsheets.

It follows the [Model Context Protocol](https://modelcontextprotocol.io), an open standard for connecting AI tools to data. The server is hosted by Volteras, so there is nothing to install or run yourself.

<Card title="Server URL" icon="link">
  ```
  https://mcp.volteras.com/mcp
  ```

  Streamable HTTP transport, authenticated with OAuth 2.1.
</Card>

<Note>
  The MCP server is available in production only, not in the [sandbox environment](/overview/sandbox-environment). It needs to be switched on for your organization first. [Contact us](https://volteras.com/contact-us) to enable it.
</Note>

## What you can do

* **Look up vehicles.** Find vehicles by make, model, connection status or when they were connected.
* **Check live state.** See state of charge, range and odometer, and when each value was last reported.
* **Find where vehicles are.** See position, speed and heading, when your connection has the `location:read` scope.
* **Spot connection problems.** List vehicles that need reconnecting or have stopped reporting.

The server is read-only. It cannot send commands to vehicles or change anything in your account.

## Get started

<CardGroup cols={2}>
  <Card title="Connect your AI client" icon="plug" href="/overview/mcp-setup">
    Set up Claude, Claude Code, VS Code or your own agent.
  </Card>

  <Card title="Authentication" icon="key" href="/overview/mcp-authentication">
    Sign in with your Volteras Portal account, or use an MCP credential for systems and agents.
  </Card>

  <Card title="Tools reference" icon="wrench" href="/overview/mcp-tools">
    The tools the server exposes, with every field and filter.
  </Card>

  <Card title="Example prompts" icon="message" href="#example-prompts">
    Things to try once you are connected.
  </Card>
</CardGroup>

## How access works

Every connection belongs to one Volteras organization and only ever sees that organization's vehicles. What it can read is limited in three ways:

1. **Who is connecting.** When you sign in as yourself, the assistant can only see what your role in the Volteras Portal allows.
2. **The connection's scopes.** `vehicles:read` lets the assistant list vehicles. `location:read` adds position data.
3. **Each vehicle's consent.** Values like the VIN or telemetry are only returned when the vehicle's connection grants access to them, just as with the [Volteras API](/overview/scopes).

See [Authentication](/overview/mcp-authentication) for the details.

## Example prompts

Once connected, try asking your assistant:

<AccordionGroup>
  <Accordion title="Fleet summary">
    ```
    Give me a summary of our connected fleet: how many vehicles we have by make and model, and how many are not fully connected.
    ```
  </Accordion>

  <Accordion title="Low charge">
    ```
    Which vehicles are below 20% state of charge? Show the make, model, charge level and when it was last reported, lowest first.
    ```
  </Accordion>

  <Accordion title="Reconnections needed">
    ```
    Which vehicles need to be reconnected? Group them by manufacturer.
    ```
  </Accordion>

  <Accordion title="Where a vehicle is">
    ```
    Where was the Model 3 with registration AB12 CDE last seen, and when?
    ```

    Needs the `location:read` scope.
  </Accordion>
</AccordionGroup>

<Tip>
  If an answer looks incomplete, ask the assistant which scopes it has. It can call the `get_me` tool to find out.
</Tip>