# Connect Claude Desktop, Cursor or VS Code

_Connect an AI assistant · 2 min read · Updated 2026-10-28 · https://lanachat.ai/guides/ai-connect-desktop-editors/_

Desktop apps and code editors read the LanaChat server from a JSON file. Find your app below, copy its block, replace `YOUR_MCP_TOKEN` with your token ([guide](https://lanachat.ai/guides/ai-create-mcp-token/)), save, and restart the app.

> Note: **Copy client configuration** in **Users › API & MCP** gives the address and your token in one `mcpServers` block with `"type": "http"`. That block is not valid for Claude Desktop (it needs `command` and `args`) nor for VS Code (it needs `servers`). Use the blocks below, and take only the address and the token from the Console.

Using Claude Code in a terminal? One command is enough ([guide](https://lanachat.ai/guides/ai-connect-claude-code/)):

```bash
claude mcp add --transport http --scope user lanachat https://api.lanachat.ai/mcp --header "Authorization: Bearer YOUR_MCP_TOKEN"
```

## Claude Desktop

Claude Desktop starts local servers only, so it reaches LanaChat through the small `mcp-remote` bridge. The bridge needs Node.js 18 or later on your computer.

1. Open Claude Desktop, then **Settings › Developer › Edit Config**. This opens the file:
  - Mac: `~/Library/Application Support/Claude/claude_desktop_config.json`
  - Windows: `%APPDATA%\Claude\claude_desktop_config.json`
2. Put this in the file (if it already has an `mcpServers` block, add only the `lanachat` part inside it):

```json
{
  "mcpServers": {
    "lanachat": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://api.lanachat.ai/mcp", "--header", "Authorization:${LANACHAT_AUTH}"],
      "env": { "LANACHAT_AUTH": "Bearer YOUR_MCP_TOKEN" }
    }
  }
}
```

Save, quit Claude Desktop completely, and open it again.

In a new chat, open the tools menu: `lanachat` is listed with its tools.

> Note: The custom connectors of the web and desktop apps that only offer an OAuth sign-in cannot send a token. Use the configuration file above.

## Cursor

1. Open the file `~/.cursor/mcp.json` (for every project) or `.cursor/mcp.json` in one project. Create it if it does not exist.
2. Add:

```json
{
  "mcpServers": {
    "lanachat": {
      "url": "https://api.lanachat.ai/mcp",
      "headers": { "Authorization": "Bearer YOUR_MCP_TOKEN" }
    }
  }
}
```

Save, then open **Settings › MCP** in Cursor and turn `lanachat` on.

## VS Code

VS Code can ask for the token when it starts the server, so the token never sits in the file.

1. In your project, create the file `.vscode/mcp.json`.
2. Add:

```json
{
  "inputs": [
    { "type": "promptString", "id": "lanachat-token", "description": "LanaChat MCP token", "password": true }
  ],
  "servers": {
    "lanachat": {
      "type": "http",
      "url": "https://api.lanachat.ai/mcp",
      "headers": { "Authorization": "Bearer ${input:lanachat-token}" }
    }
  }
}
```

Save. Press **Start** above `lanachat` in the file, paste the token when VS Code asks, and press Enter.

Open the chat in agent mode: the LanaChat tools are in the tools list.

> Warning: If a project folder goes to Git, never commit a file that holds the token. The VS Code version above holds none.

Next: [test the connection](https://lanachat.ai/guides/ai-test-connection/).

## Related

- [Create your MCP token](https://lanachat.ai/guides/ai-create-mcp-token/)
- [Test the connection](https://lanachat.ai/guides/ai-test-connection/)
- [API & MCP: connect an AI assistant](https://lanachat.ai/guides/api-mcp/)