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

# Quickstart

> Create a workspace, connect your agent over MCP, install the agent skills, and write your first document.

By the end you have a connected agent that can read your brain and write to it.

<Steps>
  <Step title="Create your workspace">
    Sign up at [tatara.app](https://www.tatara.app/signup). Signing up is free and does not need a card.

    On first sign-in you name your company. That creates your brain, the container for everything your company knows.

    <Note>
      A new brain starts empty. No starter documents, folders, or tags are created for you, so the first thing in your brain will be whatever you or your agent writes. See [plans](/account/plans) for what each plan includes.
    </Note>
  </Step>

  <Step title="Add the MCP server">
    Tatara exposes an HTTP MCP server. Add it to your agent's MCP configuration:

    ```json mcp-config.json theme={null}
    {
      "mcpServers": {
        "tatara": {
          "type": "http",
          "url": "https://www.tatara.app/api/mcp"
        }
      }
    }
    ```

    The first time your client connects, it opens a browser, asks you to sign in to Tatara, and shows a consent screen naming the client. Approving it lets that client read your documents and folders, traverse your brain's links, and create and edit documents on your behalf.

    Connected clients are listed under **Agent access** in settings, where you can disconnect any of them later.
  </Step>

  <Step title="Install the agent skills">
    The skills teach your agent the write workflow, the document formats, and how to recover when a write is rejected. Without them an agent's first write usually bounces off the format checks.

    ```bash theme={null}
    npx skills add tatara-app/tatara-skills
    ```

    That installs every skill into each agent detected on your machine. To install one skill, target specific agents, or install as a plugin, see [agent skills](/skills).
  </Step>

  <Step title="Verify, then write the first document">
    Ask your agent to list what is in the brain. It calls `list_documents` and reports what it finds. On a brand-new brain the honest answer is "nothing", which still proves the connection works.

    Then ask it to write something. A new brain has no folders yet, so the agent creates one first, then files the document into it with a title, description, and tags.

    <Tip>
      Ask the agent to check the taxonomy before it writes. `get_taxonomy` returns the folders that actually exist, the document types already in use, and your tag registry, so the agent writes into your structure instead of inventing one.
    </Tip>

    <Check>
      The document appears in the web app's sidebar, and your agent can read it back by path.
    </Check>

    If the write comes back rejected, that is normal. The rejection says what to fix, and a capable agent will correct itself and retry.
  </Step>
</Steps>

## Next steps

<Columns cols={2}>
  <Card title="Core concepts" icon="book-open" href="/concepts">
    How documents, folders, links, and tags fit together.
  </Card>

  <Card title="The MCP server" icon="wrench" href="/mcp/overview">
    What your agent can do against your brain.
  </Card>

  <Card title="Document formats" icon="table" href="/formats/overview">
    Ask for a board, a tracker, a table, or a saved view instead of plain prose.
  </Card>

  <Card title="Working in the app" icon="pen-line" href="/app/documents">
    Writing, organizing, and searching your brain by hand.
  </Card>
</Columns>