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

# Get started

> Install the GRID CLI, connect your account, and run your first session

Everything in GRID starts from the `grid` CLI — an interactive shell that
onboards robots, starts simulator and robot sessions, and deploys programs.

## 1. Install the CLI

```bash theme={null}
curl -fsSL https://generalrobotics.dev/install.sh | bash
```

Your organization's **GRID admin portal** shows this same command on its
Welcome screen, pre-filled with your cluster:

```bash theme={null}
curl -fsSL https://generalrobotics.dev/install.sh | bash -s -- \
  --cluster-name <your-org> --cluster-url <your-cluster-url>
```

Prefer the portal variant — it registers your cluster so `login` connects to
the right place immediately. (Windows support is coming soon; use WSL in the
meantime.)

The installer verifies a signed binary, installs it to `~/.grid/bin/grid`,
adds it to your `PATH`, and offers the GRID Studio VS Code extension. Open a
new terminal and run:

```bash theme={null}
grid
```

## 2. Log in

Inside the shell, authenticate with your browser:

```text theme={null}
GRID # login
```

Your tokens are stored per cluster; `switch` changes the active cluster if
your organization has more than one.

## 3. Get your Cortex API key

Cloud model inference (detection, depth, grasping, VLAs) is authenticated by a
separate runtime key. Generate it in the **admin portal** (the same site that
showed your install command — ask your org admin for its URL if you don't
have it), and paste it when `robot add` asks — the CLI persists it for every
later session.

## 4. Onboard a robot

```text theme={null}
GRID # robot add
```

The wizard walks you through connecting your robot to the GRID network. Once
added, it shows up for everyone in your organization:

```text theme={null}
GRID # robot list
```

## 5. Start working

```text theme={null}
GRID # session start
```

Pick **sim** for a cloud simulator session (opens a browser workspace with a
live scene), or **robot** to open a VS Code workspace wired to your physical
robot — a notebook with a live `robot` connection, ready for code.

## Where next

* Watch a simulator come alive, no hardware needed: [Your first session](/get-started/first-session)
* Write and deploy your first robot program: [Your first skill](/get-started/first-skill)
* Every command: the [CLI reference](/cli/reference)
* Call cloud models from Python: [grid-cortex-client](/python-api/grid-cortex-client/overview)
* The hosted model catalog: [AI models](/models/overview)
* Something wrong? `diagnostics` bundles redacted logs for support, and
  `error` recalls the last failure in full.


## Related topics

- [Overview](/models/overview.md)
- [GRID Cortex Client](/models/cortex.md)
- [Your first session](/get-started/first-session.md)
- [Your first skill](/get-started/first-skill.md)
- [The Intelligence Grid for Physical AI](/introduction.md)