Your agent
Connect your own agent to nextanalytica's MCP server and ask about your site in plain words, instead of reading a dashboard.
Updated
nextanalytica has no dashboard to read, and that's the point. Your own agent connects to our MCP server, reads your project's visits, sources and team context, and answers in plain words. It works with Claude Code, the Claude app, Codex, Cursor, VS Code, Windsurf, Zed and any client that speaks remote MCP.
How do I connect my agent?
The server's address is the same for every client:
https://mcp.nextanalytica.app/mcpWe recommend the fastest way: click Onboard your agent in the dashboard and paste the message into your agent. An agent that runs commands (Claude Code, Codex, Cursor) connects itself and tells you what it sees; in the Claude app, it tells you the clicks to add the connector. Or pick your client:
| Client | How |
|---|---|
| Claude Code | One command in the terminal |
| Claude app | A custom connector in Settings |
| Codex | One command |
| Cursor | .cursor/mcp.json |
| VS Code | .vscode/mcp.json |
| Windsurf | mcp_config.json |
| Zed | Zed's settings |
| Any MCP client | The address, with OAuth or a key |
Each person signs in as themselves, in the browser or with an agent key. One connection covers every project you can see, not one per site.
What can I ask it?
"How did shop.example do this week?": the week against the week before, with the three biggest changes and where they came from.
"Why did visits drop on Tuesday?": the channel, page or campaign behind the drop, with the numbers.
"Add nextanalytica to my Next.js site.": your agent takes the project's tag line, puts it in your layout, and asks before committing.
There's a lot more in Use cases.
What will it never do?
- Invent a number. When the data can't answer, it says so.
- Change your website through nextanalytica. Our tools never change a site: your agent changes code only in your repository with your approval, or through a connector your team added.
- See more than you can see.
We run no AI model on your data. The thinking happens in your agent, on your plan.
Go deeper
- Use cases: questions that work, by task, and multi-step recipes.
- MCP tools: every tool, its arguments and example prompts.
- FAQ: cost, clients, keys, data, safety.
- Team access: who can connect, and how access ends.