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

# Quickstart

> Get Arden running locally.

## Prerequisites

* Python 3.13+ with [uv](https://docs.astral.sh/uv/)
* [Bun](https://bun.sh) for the desktop app when running from source
* An LLM provider: Anthropic, OpenAI, Google, OpenRouter, custom OpenAI-compatible endpoint, or OpenAI account sign-in

## Install

Arden currently runs from a source checkout:

```bash theme={null}
git clone https://github.com/esceptico/arden.git arden
cd arden
cp .env.example .env
just install
```

## Set up

<Steps>
  ### Set your LLM provider

  Set one provider key in `.env`, or configure a provider in the desktop app after connecting.

  ```bash theme={null}
  # Edit .env and uncomment one:
  ANTHROPIC_API_KEY=sk-ant-...
  # OPENAI_API_KEY=sk-...
  # GEMINI_API_KEY=...
  ```

  If you want to use an OpenAI subscription/account instead of an API key, skip the key export and choose **OpenAI Codex** in provider onboarding. Arden opens a browser sign-in and stores refreshable OAuth tokens in `~/.arden/openai-codex-auth.json`.

  ### Start the server

  <Tabs>
    <Tab title="From source">
      ```bash theme={null}
      just server
      ```
    </Tab>

    <Tab title="Docker">
      ```bash theme={null}
      docker compose up --build
      ```
    </Tab>
  </Tabs>

  Compose publishes the server only on `127.0.0.1` and persists Arden state in
  the `arden-data` volume.

  On first run, Arden generates an API key and prints it:

  ```
  Your API key: nBx7k2...
  Enter this in a desktop client to connect. It won't be shown again.
  ```

  <Warning>Copy this key – it is only shown once. Use `uv run --project apps/server arden-server serve --reset-key` to generate a new one.</Warning>

  ### Start the desktop app

  Desktop currently requires a source checkout. In a second terminal:

  ```bash theme={null}
  just desktop
  ```

  Paste your API key in the connection screen. If no LLM provider is configured, provider onboarding appears so you can connect a provider and choose defaults.

  ### Connect integrations

  Open **Settings → Integrations** to connect Gmail, Google Calendar, Google
  Drive, or Slack. Use **Settings → MCP** for MCP servers.

  <img src="https://mintcdn.com/ntrp/yXMP7sfCaHIwY_fB/images/settings-integrations.png?fit=max&auto=format&n=yXMP7sfCaHIwY_fB&q=85&s=3add196e7c30c92ed5f047288a10ec04" alt="Connected integrations" width="2640" height="1700" data-path="images/settings-integrations.png" />
</Steps>

## What's next?

<CardGroup cols={2}>
  <Card title="Desktop Guide" icon="desktop" href="/guides/desktop">
    Learn the app surfaces: chat, approvals, settings, memory, and traces.
  </Card>

  <Card title="Configuration" icon="gear" href="/configuration">
    Customize models, integrations, and runtime settings.
  </Card>

  <Card title="Memory" icon="brain" href="/guides/memory">
    Understand how Arden remembers things.
  </Card>

  <Card title="Connect Gmail" icon="envelope" href="/integrations/gmail">
    Let Arden read and send emails.
  </Card>
</CardGroup>