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

# Connect an AI assistant

> Add Alura to Claude, Claude Code, ChatGPT or Cursor with its MCP server.

Alura's MCP server is at:

```text theme={null}
https://mcp.alura.io/mcp
```

Add it to your assistant, then sign in with your Alura account when it asks. The approval screen shows what the assistant can do (read research, read and change saved research, read usage) and which workspace it starts in. See [Authentication](/api-reference/authentication#mcp-connections) for how workspaces work.

<Tabs>
  <Tab title="Claude.ai">
    1. In Claude, open **Settings**, then **Connectors**.
    2. Choose **Add custom connector**, name it `Alura`, and paste the server URL.
    3. Choose **Connect**, sign in to Alura and approve.

    In a chat, turn Alura on from the tools menu and ask, for example, "Which personalized mugs under \$30 sell best on Etsy?"
  </Tab>

  <Tab title="Claude Code">
    Add the server, then sign in from Claude Code:

    ```bash theme={null}
    claude mcp add --transport http alura https://mcp.alura.io/mcp
    ```

    Start Claude Code, run `/mcp`, pick **alura** and choose **Authenticate**. Your browser opens to sign in to Alura and approve.
  </Tab>

  <Tab title="ChatGPT">
    1. In ChatGPT, open **Settings**, then **Apps & Connectors**, then **Advanced settings**, and turn on **Developer mode**.
    2. Back in **Apps & Connectors**, choose **Create**. Name it `Alura`, paste the server URL and pick **OAuth** for authentication.
    3. Sign in to Alura and approve.

    Once Alura is listed in ChatGPT's app directory, you can add it from there instead.
  </Tab>

  <Tab title="Cursor">
    Add Alura to `~/.cursor/mcp.json` (all projects) or `.cursor/mcp.json` (one project):

    ```json theme={null}
    {
      "mcpServers": {
        "alura": {
          "url": "https://mcp.alura.io/mcp"
        }
      }
    }
    ```

    Cursor lists Alura under **Settings**, then **MCP**, with a **Needs login** button. Choose it to sign in to Alura and approve.
  </Tab>
</Tabs>

## Clients without sign-in

A client that can't do OAuth can send an API key instead. The key's workspace is used, and it can't switch workspace.

```json theme={null}
{
  "mcpServers": {
    "alura": {
      "url": "https://mcp.alura.io/mcp",
      "headers": { "Authorization": "Bearer alura_sk_..." }
    }
  }
}
```

## What the assistant can do

The tools cover the research, saved research and usage endpoints, at the same prices: searching and opening products and shops, reviews, keywords, categories, saved research and usage. Each tool's description ends with its cost, so the assistant knows it before calling. Every result carries what it cost in `_meta.credits_charged`.

Assistants see a little less than the API. Etsy's API has no flag for mature listings, so Alura counts products in Etsy's Sexual Wellness category, and every category under it, as mature. Through an assistant, searches, trending lists, similar products, a shop's products and reviews, keyword top results and the category browser leave them out, and so do shops whose active products are all mature. Asking for one by ID or Etsy link answers `not_found`, and in a saved list it shows as `unavailable`. The API with an API key returns them as before.

Review text and shop announcements are written by other people. Alura leaves them out of the text your assistant reads and passes them only as structured data. Your assistant should never follow instructions found in them.

To stop a connection, revoke it in platform.alura.io. A workspace owner or admin can also remove their workspace from anyone's connection.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.