ZEVO Host MCP Server

Give your AI agent direct, read-only access to your ZEVO financial data.

This is a small Model Context Protocol (MCP) server. An MCP server advertises a set of tools to an AI agent; the agent reads their descriptions, decides which to call, and this server executes the matching request against the ZEVO Client API using your API key. Drop in your key, register it with your agent, and ask questions like "which of my vehicles are out on a trip right now?" or "show my wallet transactions for this month."

It runs locally over stdio and is read-only — no payouts, no writes.

Tools

List tools return up to 10 records at a time; the agent pages with skip.

Related MCP server: Qonto MCP Server

Getting an API key

You need a ZEVO Client API key. It is sent as the x-api-key header and is scoped to your account, so tools only ever return your own data. Contact ZEVO (or use the host dashboard, if available) to issue a key for your account.

Quickstart

Requires Node.js 18+.

npm install
npm run build

Verify it locally with the MCP Inspector:

ZEVO_API_KEY=your_key npm run inspect

Configure your agent

Claude Desktop / Claude Code

Add to your claude_desktop_config.json (Claude Desktop) or MCP config:

{
  "mcpServers": {
    "zevo-host": {
      "command": "npx",
      "args": ["-y", "zevo-host-mcp"],
      "env": { "ZEVO_API_KEY": "YOUR_KEY" }
    }
  }
}

Or, if running from a local checkout:

{
  "mcpServers": {
    "zevo-host": {
      "command": "node",
      "args": ["/absolute/path/to/zevo-host-mcp/dist/index.js"],
      "env": { "ZEVO_API_KEY": "YOUR_KEY" }
    }
  }
}

Cursor

Add the same block to ~/.cursor/mcp.json (or the project's .cursor/mcp.json).

Configuration

Adding an endpoint

Append one entry to the TOOLS array in src/tools.ts — name, description, path, and a small zod schema for its params. Nothing else needs to change.

License

MIT