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

# Lexmount Cloud Browser

> Use Lexmount cloud browser for large-scale automated evaluation

Lexmount provides cloud-based browser instances for running browser automation tasks. Supports session persistence, login state management, and large-scale parallel execution.

## Service Endpoints

Lexmount offers two production regions. Choose the one closest to where your evaluation is running:

| Environment                 | Console                                              | API base URL               |
| --------------------------- | ---------------------------------------------------- | -------------------------- |
| Production (mainland China) | [browser.lexmount.cn](https://browser.lexmount.cn)   | `https://api.lexmount.cn`  |
| Production (international)  | [browser.lexmount.com](https://browser.lexmount.com) | `https://api.lexmount.com` |

The API base URL is optional — the SDK defaults to the mainland China endpoint. Override it via `LEXMOUNT_BASE_URL` in `.env` or `lexmount_base_url` in `config.yaml` when running from outside mainland China.

## Prerequisites

<Steps>
  <Step title="Get API Key">
    Register at the Lexmount console ([browser.lexmount.cn](https://browser.lexmount.cn) for mainland China, [browser.lexmount.com](https://browser.lexmount.com) for international) and apply for an API Key.
  </Step>

  <Step title="Get Project ID">
    Create a project in the same console to get the Project ID.
  </Step>
</Steps>

## Configuration

Add credentials to `.env`:

```bash theme={null}
LEXMOUNT_API_KEY=your_api_key
LEXMOUNT_PROJECT_ID=your_project_id
```

Copy the example config (`cp config.example.yaml config.yaml`), then configure the cloud browser for the target agent in the root `config.yaml`:

```yaml theme={null}
# config.yaml
agents:
  browser-use:
    browser:
      browser_id: lexmount
      lexmount_browser_mode: normal   # normal | light
      lexmount_api_key: $LEXMOUNT_API_KEY
      lexmount_project_id: $LEXMOUNT_PROJECT_ID
```

## Login Contexts

For tasks requiring login, use the `bubench login` subcommand to create a **persistent Lexmount context** that stores the login state. At evaluation time the runtime auto-forks this context based on each task's `target_website` — one login serves many concurrent runs.

### Step 1: Save a Login

```bash theme={null}
# Xiaohongshu
bubench login add xiaohongshu

# Taobao
bubench login add taobao

# Custom URL (sites not in the built-in registry)
bubench login add my_site --website https://example.com/login
```

The command prints an **Inspect URL** — open it in your local browser, complete the login, then return to the terminal and press **Enter** to persist cookies into the context.

<Tip>
  Nearly 100 common sites are built in (e-commerce, social, video, office/collab, etc.). Run `bubench login add help` to list them all, or edit `browseruse_bench/browsers/login_sites.yaml` to add your own.
</Tip>

### Step 2: Login Flow

<Steps>
  <Step title="Run the command">
    `bubench login add <site>` creates the context, opens a remote session, and prints an Inspect URL
  </Step>

  <Step title="Log in via the browser">
    Open the Inspect URL locally — the remote Chrome is already navigated to the target site. Log in normally
  </Step>

  <Step title="Confirm & save">
    Press **Enter** in the terminal. The session closes cleanly and cookies persist into the base context
  </Step>
</Steps>

The mapping is recorded in `browser_data/login_contexts/index.json`.

### Step 3: Other Commands

```bash theme={null}
# List all saved login contexts
bubench login list

# Remove local mapping only (keep remote context)
bubench login remove xiaohongshu

# Also delete the remote Lexmount context
bubench login remove xiaohongshu --delete-remote
```

## Running Tasks with Login

Any task with `login_required: true` whose `target_website` resolves to a saved site will auto-mount the matching context — no extra flags needed:

<Steps>
  <Step title="Ensure login exists">
    Run `bubench login list` to confirm the target site is saved
  </Step>

  <Step title="Use Normal mode">
    ```yaml theme={null}
    # config.yaml
    agents:
      browser-use:
        browser:
          browser_id: lexmount
          lexmount_browser_mode: normal
    ```
  </Step>

  <Step title="Run tasks">
    ```bash theme={null}
    bubench run \
      --agent browser-use \
      --data LexBench-Browser \
      --mode first_n --count 5
    ```
  </Step>
</Steps>

## Advantages

<CardGroup cols={2}>
  <Card title="No Local Browser" icon="laptop-slash">
    Runs in server environments without Chrome installed
  </Card>

  <Card title="Session Persistence" icon="database">
    Login state persists across tasks, no repeated logins
  </Card>

  <Card title="Parallel Execution" icon="layer-group">
    Supports multiple browser instances running in parallel
  </Card>

  <Card title="Stable & Reliable" icon="shield-check">
    Cloud-isolated environment avoids local resource conflicts
  </Card>
</CardGroup>

## Notes

<Warning>
  * Cloud browser usage incurs costs — use responsibly
  * Normal mode sessions have an expiry time — check login state regularly
  * Do not expose your API Key in public environments
</Warning>