# MCP quick start

MCP gives an AI agent tools for reading analytics, researching videos, and working with scenarios. The connection uses OAuth, so its configuration needs no secrets.

## Server address

```text
https://api.viralmaxing.com/api/mcp
```

The server uses Streamable HTTP.

## Claude Code

Run this command in the project directory:

```bash
claude mcp add --transport http --scope project viralmaxing https://api.viralmaxing.com/api/mcp
```

Open `/mcp` in the session and start sign-in for `viralmaxing`. Choose a workspace in the browser and approve access.

## Claude, Cursor, and other clients

If the client accepts JSON configuration, add the server without custom headers or keys:

```json
{
  "mcpServers": {
    "viralmaxing": {
      "type": "http",
      "url": "https://api.viralmaxing.com/api/mcp"
    }
  }
}
```

In Claude, open connector settings and add a custom connector at the same address. In Cursor, use MCP settings. The client opens Viralmaxing sign-in after saving the server.

[Settings → API keys](/settings/api) contains a ready-to-use agent prompt. It asks the agent to add the server, complete OAuth, and check the tool list.

## First request

Ask the agent to call `list_workspaces`, then `list_my_accounts`. Both calls only read data and spend no energy.

Continue with [connection verification](/developers/mcp/verification) to confirm that the agent is using the intended workspace.