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

# Installation

> Install the Browser-act CLI, configure the entry Skill, and verify the runtime.

Install Browser-act in two parts:

1. The **Browser-act CLI**, which performs browser automation.
2. The **entry Skill**, which tells your AI agent when and how to call the CLI.

<Columns cols={2}>
  <Card title="1. Add the entry Skill" icon="sparkles">
    Let your agent discover Browser-act and learn that `get-skills` must be loaded before browser work.
  </Card>

  <Card title="2. Verify the CLI" icon="terminal">
    Install `browser-act-cli`, check the version, and confirm the runtime can execute commands.
  </Card>
</Columns>

## Agent installation

For agents that support skills, the shortest path is to ask the agent to install Browser-act for you:

```text theme={null}
Install browser-act.
Skill URL: https://github.com/browser-act/skills/tree/main/browser-act
After installation, verify that the CLI is available.
```

The agent should install the CLI, add the Skill file to its skill directory, and run a basic availability check.

## Manual CLI installation

Install the CLI with `uv` and Python 3.12:

```bash theme={null}
uv tool install browser-act-cli --python 3.12
```

Verify the command:

```bash theme={null}
browser-act --version
```

## Optional: API key

> \[!NOTE]
> Browser-act can run basic Chrome and chrome-direct browser automation without an API key. An API key unlocks hosted capabilities: `stealth` browsers, `stealth-extract`, dynamic proxy rotation, and `solve-captcha`.

Start the login flow:

```bash theme={null}
browser-act auth login
browser-act auth poll
```

Or set a key directly:

```bash theme={null}
browser-act auth set <your_api_key>
```

## Upgrade

Upgrade the CLI with:

```bash theme={null}
uv tool upgrade browser-act-cli
```

> \[!TIP]
> The Skill layer can detect version mismatches between the installed Skill and the CLI. When that happens, `get-skills` returns upgrade guidance for the agent.

## Requirements

| Requirement        | Notes                                                   |
| ------------------ | ------------------------------------------------------- |
| Python 3.12+       | Required by the CLI package                             |
| Chrome or Chromium | Required for `chrome` and `chrome-direct` browser types |
| API key            | Required only for `stealth` browser features            |

## Troubleshooting

### Command not found after installation

> \[!WARNING]
> If `browser-act` is not found after installation, make sure the `uv` tool directory is in your shell `PATH`.

```bash theme={null}
uv tool dir
```

Add the returned directory to `PATH`, then open a new terminal.

### Send diagnostics

Upload diagnostic logs:

```bash theme={null}
browser-act report-log
```

Send product feedback:

```bash theme={null}
browser-act feedback "Describe the issue or suggestion"
```

## What is next

<Columns cols={2}>
  <Card title="Continue to Quick Start" icon="rocket" href="/agent-cli/quick-start" cta="Run the first task">
    Run a content extraction or a browser session in about a minute.
  </Card>

  <Card title="Learn the Skill System" icon="layers" href="/agent-cli/skill-system" cta="Understand the runtime">
    See how the entry Skill and `get-skills` keep agent instructions current.
  </Card>
</Columns>