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

# 浏览器沙箱

> 一个安全的浏览器沙箱，代理可在其中与 Web 交互。

<Info>
  对于代理工作流，请使用 [Interact](/zh/features/interact)。Interact 是 CLI/MCP 的受支持方案，可在抓取后通过提示词或代码驱动；MCP 也支持直接从 URL 打开。
</Info>

| Surface  | 用途                                                           | 入口                                                                                   | 代理入口                        |
| -------- | ------------------------------------------------------------ | ------------------------------------------------------------------------------------ | --------------------------- |
| 浏览器沙箱    | 面向需要沙箱、CDP URL、实时视图或持久会话生命周期的 API/SDK 用户的独立浏览器会话             | `POST /v2/interact`                                                                  | API 和 SDKs；隐藏的 CLI 浏览器命令为旧版 |
| Interact | 对已抓取的页面执行操作；MCP 也可通过 `firecrawl_interact` 的 URL 模式直接从 URL 打开 | `POST /v2/scrape/{scrapeId}/interact`、抓取后的 CLI `interact`，或 MCP `firecrawl_interact` | 推荐用于 CLI/MCP 代理工作流          |

Firecrawl 浏览器沙箱 为 API 和 SDK 用户提供一个安全的浏览器环境，让代理能够与 Web 交互。可以填写表单、点击按钮、进行身份验证等。
无需本地配置，无需安装 Chromium，也不存在驱动兼容性问题。代理浏览器和 Playwright 已预先安装。

可通过 [API](/zh/api-reference/endpoint/browser-create)、[Node SDK](/zh/sdks/node#browser)、[Python SDK](/zh/sdks/python#browser) 和 [Vercel AI SDK](/zh/developer-guides/llm-sdks-and-frameworks/vercel-ai-sdk) 使用。隐藏的 `firecrawl browser` CLI 命令为旧版；CLI 和 MCP 代理流程应改用 scrape + interact。

要为 AI 编码代理 (Claude Code、Codex、Open Code、Cursor 等) 添加 Interact 支持，请安装 Firecrawl skill：

```bash theme={null}
npx -y firecrawl-cli@latest init --all --browser
```

每个会话都在隔离的一次性或持久化沙箱中运行，并且可在无需管理基础设施的情况下进行扩展。

<div id="quick-start">
  ## 快速开始
</div>

创建会话，执行代码，然后关闭它：

<CodeGroup>
  ```js Node theme={null}
  // npm install firecrawl
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  // 1. Launch a session
  const session = await firecrawl.browser();
  console.log(session.cdpUrl); // wss://cdp-proxy.firecrawl.dev/cdp/...

  // 2. Execute code
  const result = await firecrawl.browserExecute(session.id, {
    code: `
      await page.goto("https://news.ycombinator.com");
      const title = await page.title();
      console.log(title);
    `,
    language: "node",
  });
  console.log(result.result); // "Hacker News"

  // 3. Close
  await firecrawl.deleteBrowser(session.id);
  ```

  ```python Python theme={null}
  # pip install firecrawl
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  # 1. 启动会话
  session = app.browser()
  print(session.cdp_url)  # wss://cdp-proxy.firecrawl.dev/cdp/...

  # 2. Execute code
  result = app.browser_execute(
      session.id,
      code='await page.goto("https://news.ycombinator.com")\ntitle = await page.title()\nprint(title)',
      language="python",
  )
  print(result.result)  # "Hacker News"

  # 3. Close
  app.delete_browser(session.id)
  ```

  ```bash CLI theme={null}
  # Install the Firecrawl CLI
  npm install -g firecrawl-cli

  # 简写方式 - 自动启动会话,无需 "execute"
  firecrawl browser "open https://news.ycombinator.com"
  firecrawl browser "snapshot"
  firecrawl browser "scrape"

  # Close when done
  firecrawl browser close
  ```

  ```bash cURL theme={null}
  # 1. 启动会话
  curl -X POST "https://api.firecrawl.dev/v2/interact" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json"

  # 2. 执行代码
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "code": "await page.goto(\"https://news.ycombinator.com\")\ntitle = await page.title()\nprint(title)"
    }'

  # 3. 关闭会话
  curl -X DELETE "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY"
  ```
</CodeGroup>

* **无需安装驱动** - 无需 Chromium 二进制文件，无需 `playwright install`，也没有驱动兼容性问题
* **Python、JavaScript 和 Bash** - 通过 API、CLI 或 SDK 发送代码并获取结果。三种语言都会在远程沙盒中运行
* **agent-browser** - 预装的 CLI，内置 60+ 条命令。AI 代理只需编写简单的 Bash 命令，无需编写 Playwright 代码
* **已加载 Playwright** - Playwright 已预装在沙盒中。代理如果愿意，也可以编写 Playwright 代码。
* **CDP 访问** - 当你需要完全控制时，可通过 WebSocket 连接你自己的 Playwright 实例
* **实时视图** - 通过可嵌入的流 URL 实时查看会话
* **交互式实时视图** - 通过可嵌入的交互式流，让用户直接与浏览器交互

<div id="launch-a-session">
  ## 启动一个会话
</div>

返回会话 ID、CDP URL 和实时视图 URL。

<CodeGroup>
  ```js Node theme={null}
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  const session = await firecrawl.browser({
    ttl: 120,
    activityTtl: 60,
  });

  console.log(session.id);
  console.log(session.cdpUrl);      // wss://cdp-proxy.firecrawl.dev/cdp/...
  console.log(session.liveViewUrl); // https://liveview.firecrawl.dev/...
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  session = app.browser(
      ttl=120,
      activity_ttl=60,
  )

  print(session.id)
  print(session.cdp_url)        # wss://cdp-proxy.firecrawl.dev/cdp/...
  print(session.live_view_url)  # https://liveview.firecrawl.dev/...
  ```

  ```bash CLI theme={null}
  # 启动实时视图并自定义 TTL
  firecrawl browser launch-session --stream --ttl 120 --ttl-inactivity 60

  # 启动并将会话信息保存至文件
  firecrawl browser launch-session -o session.json --json
  ```

  ```bash cURL theme={null}
  curl -X POST "https://api.firecrawl.dev/v2/interact" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "ttl": 120,
      "activityTtl": 60
    }'
  ```
</CodeGroup>

```json Response theme={null}
{
  "success": true,
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "cdpUrl": "wss://browser.firecrawl.dev/cdp/550e8400...?token=abc123...",
  "liveViewUrl": "https://liveview.firecrawl.dev/...",
  "interactiveLiveViewUrl": "https://liveview.firecrawl.dev/...",
  "expiresAt": "2025-01-15T10:40:00Z"
}
```

<div id="execute-code">
  ## 执行代码
</div>

在会话中运行 Python、JavaScript 或 bash 代码。输出通过 `stdout` 返回；对于 Node.js，最后一个表达式的值也可在 `result` 中获得。

<CodeGroup>
  ```js Node theme={null}
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  const result = await firecrawl.browserExecute("YOUR_SESSION_ID", {
    code: 'await page.goto("https://example.com"); const title = await page.title(); console.log(title);',
    language: "node",
  });

  console.log(result);
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  result = app.browser_execute(
      "YOUR_SESSION_ID",
      code='await page.goto("https://example.com")\ntitle = await page.title()\nprint(title)',
      language="python",
  )

  print(result)
  ```

  ```bash CLI theme={null}
  # agent-browser 命令(默认情况下会自动添加 "agent-browser" 前缀)
  firecrawl browser execute "open https://example.com"
  firecrawl browser execute "snapshot"
  firecrawl browser execute "scrape"

  # Execute Playwright Python code
  firecrawl browser execute --python 'await page.goto("https://example.com")
  print(await page.title())'

  # Execute Playwright JavaScript code
  firecrawl browser execute --node 'await page.goto("https://example.com"); document.title'

  # Execute arbitrary bash in the sandbox
  firecrawl browser execute --bash 'ls /tmp'

  # Target a specific session
  firecrawl browser execute --session <id> "snapshot"
  ```

  ```bash cURL theme={null}
  # Execute Playwright Python code
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "code": "await page.goto(\"https://example.com\")\ntitle = await page.title()\nprint(title)",
      "language": "python"
    }'

  # 执行 Playwright JavaScript 代码
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "code": "await page.goto(\"https://example.com\"); const title = await page.title(); console.log(title);",
      "language": "node"
    }'
  ```

  ```bash cURL (Bash / agent-browser) theme={null}
  # Navigate to a page
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{ "code": "agent-browser open https://example.com", "language": "bash" }'

  # Get accessibility snapshot with @ref IDs
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{ "code": "agent-browser snapshot", "language": "bash" }'

  # 使用快照中的 @ref ID 进行交互
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{ "code": "agent-browser click @e5", "language": "bash" }'

  # Extract page content as markdown
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{ "code": "agent-browser scrape", "language": "bash" }'
  ```
</CodeGroup>

```json Response theme={null}
{
  "success": true,
  "stdout": "",
  "result": "Example Domain",
  "stderr": "",
  "exitCode": 0,
  "killed": false
}
```

<div id="handling-file-downloads">
  ### 处理文件下载
</div>

在会话中下载的文件可以捕获并以 base64 形式返回。可通过 execute 端点使用 Playwright 的下载 API：

<CodeGroup>
  ```python Python theme={null}
  import base64

  async with page.expect_download() as download_info:
      await page.click('a#download-link')  # 点击触发下载的元素

  download = download_info.value
  path = await download.path()

  # 可选：保存到指定路径
  # await download.save_as('/tmp/myfile.pdf')

  # 读取文件内容并以 base64 输出
  with open(path, "rb") as f:
      content = base64.b64encode(f.read()).decode()
      print(content)
  ```

  ```javascript Node theme={null}
  // 从链接元素获取下载 URL
  const href = await page.getAttribute('a#download-link', 'href');

  // 在浏览器上下文中获取文件并转换为 base64
  const b64 = await page.evaluate(async (url) => {
    const resp = await fetch(url);
    const blob = await resp.blob();
    return new Promise((resolve) => {
      const reader = new FileReader();
      reader.onloadend = () => resolve(reader.result.split(',')[1]);
      reader.readAsDataURL(blob);
    });
  }, href);

  process.stdout.write(b64);
  ```
</CodeGroup>

<Note>
  沙箱文件系统是临时的——会话结束后，已下载的文件就会丢失。若要持久保存文件，请在会话期间读取其内容并保存到你自己的存储中。持久化配置文件会保留浏览器状态 (cookies、localStorage) ，但不会保留磁盘上的文件。
</Note>

<div id="agent-browser-bash-mode">
  ## agent-browser (Bash 模式)
</div>

[agent-browser](https://github.com/vercel-labs/agent-browser) 是一个无头浏览器 CLI，已预装在每个沙箱中。代理无需编写 Playwright 代码，只需发送简单的 bash 命令。CLI 会自动注入 `--cdp`，使 agent-browser 自动连接到你的当前会话。

<Note>
  下面的 `firecrawl browser` CLI 示例适用于旧版 浏览器沙箱 会话。对于 CLI/MCP 代理工作流，建议优先使用 `firecrawl interact` 或 MCP `firecrawl_interact` 工具。
</Note>

<div id="shorthand">
  ### 简写
</div>

这是使用 browser 的最快方式。简写和 `execute` 都会自动将命令发送给 agent-browser。简写只是跳过了 `execute`，并在需要时自动启动会话：

```bash theme={null}
firecrawl browser "open https://example.com"
firecrawl browser "snapshot"
firecrawl browser "click @e5"
```

<div id="cli">
  ### CLI
</div>

显式用法使用 `execute` 命令。命令会自动发送到 agent-browser —— 无需手动输入 `agent-browser` 或使用 `--bash`：

<CodeGroup>
  ```bash 导航 & 快照 theme={null}
  firecrawl browser execute "open https://example.com"
  firecrawl browser execute "snapshot"
  ```

  ```bash 交互 theme={null}
  firecrawl browser execute "click @e5"
  firecrawl browser execute "fill @e3 'search query'"
  firecrawl browser execute "scrape"
  ```
</CodeGroup>

<div id="api-sdk">
  ### API 与 SDK
</div>

使用 `language: "bash"`，通过 API 或 SDK 执行 agent-browser 命令：

<CodeGroup>
  ```bash cURL theme={null}
  curl -X POST "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID/execute" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "code": "agent-browser snapshot",
      "language": "bash"
    }'
  ```

  ```javascript Node theme={null}
  const result = await app.browserExecute(sessionId, {
    code: "agent-browser snapshot",
    language: "bash",
  });
  ```

  ```python Python theme={null}
  result = app.browser_execute(
      session_id,
      code="agent-browser snapshot",
      language="bash",
  )
  ```
</CodeGroup>

<div id="session-management">
  ## 会话管理
</div>

<div id="persistent-sessions">
  ### 持久化会话
</div>

默认情况下，每个浏览器会话都会从全新环境开始。通过 `profile`，你可以在会话之间保存并复用浏览器状态。这对于保持登录状态和保留偏好设置非常有用。

若要保存或选择某个持久化配置文件，请在创建会话时使用 `profile` 参数。

<CodeGroup>
  ```js Node theme={null}
  const session = await firecrawl.browser({
    ttl: 600,
    profile: {
      name: "my-profile",
      saveChanges: true,
    },
  });
  ```

  ```python Python theme={null}
  session = app.browser(
      ttl=600,
      profile={
          "name": "my-profile",
          "save_changes": True,
      },
  )
  ```

  ```bash cURL theme={null}
  curl -X POST "https://api.firecrawl.dev/v2/interact" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
    -H "Content-Type: application/json" \
    -d '{
      "ttl": 600,
      "profile": {
        "name": "my-profile",
        "saveChanges": true
      }
    }'
  ```

  ```bash CLI theme={null}
  # 使用配置文件启动（默认保存更改）
  firecrawl browser launch-session --profile my-profile

  # 以只读模式使用配置文件启动
  firecrawl browser launch-session --profile my-profile --no-save-changes

  # 简写：使用配置文件启动并一步执行
  firecrawl browser --profile my-profile "open https://example.com"
  ```
</CodeGroup>

| 参数            | 默认值    | 描述                                                                             |
| ------------- | ------ | ------------------------------------------------------------------------------ |
| `name`        | —      | 持久化配置文件的名称。使用相同名称的会话会共享存储。                                                     |
| `saveChanges` | `true` | 当为 `true` 时，浏览器状态会在关闭时保存回该配置文件。将其设置为 `false` 可在不写入的情况下加载已有数据——在需要多个并发只读访问时很有用。 |

<Note>
  同一时间只有一个会话可以向某个配置文件保存数据。如果已有其他会话在保存，你会收到 `409` 错误。你仍然可以以 `saveChanges: false` 的方式打开同一配置文件，或者稍后重试。
</Note>

浏览器会话状态只有在会话关闭时才会保存。因此，我们建议在使用完成后关闭浏览器会话，以便后续复用。会话一旦关闭，其会话 ID 将不再有效——你无法再次使用它。请改为使用相同的配置文件名称创建一个新会话，并使用响应中返回的新会话 ID。要保存并关闭会话：

<CodeGroup>
  ```js Node theme={null}
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  await firecrawl.deleteBrowser("YOUR_SESSION_ID");
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  app.delete_browser("YOUR_SESSION_ID")
  ```

  ```bash CLI theme={null}
  # 关闭当前会话
  firecrawl browser close

  # 关闭指定会话
  firecrawl browser close --session <id>
  ```

  ```bash cURL theme={null}
  curl -X DELETE "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY"
  ```
</CodeGroup>

<div id="list-sessions">
  ### 列出会话
</div>

<CodeGroup>
  ```js Node theme={null}
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  const { sessions } = await firecrawl.listBrowsers();
  console.log(sessions);

  // 按状态过滤
  const { sessions: active } = await firecrawl.listBrowsers({ status: "active" });
  console.log(active);
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  response = app.list_browsers()
  print(response.sessions)
  ```

  ```bash CLI theme={null}
  firecrawl browser list
  firecrawl browser list active
  ```

  ```bash cURL theme={null}
  curl -X GET "https://api.firecrawl.dev/v2/interact" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY"

  # 按状态过滤
  curl -X GET "https://api.firecrawl.dev/v2/interact?status=active" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY"
  ```
</CodeGroup>

```json Response theme={null}
{
  "success": true,
  "sessions": [
    {
      "id": "550e8400-e29b-41d4-a716-446655440000",
      "status": "active",
      "cdpUrl": "wss://browser.firecrawl.dev/cdp/550e8400...?token=abc123...",
      "liveViewUrl": "https://liveview.firecrawl.dev/...",
      "interactiveLiveViewUrl": "https://liveview.firecrawl.dev/...",
      "createdAt": "2025-01-15T10:30:00Z",
      "lastActivity": "2025-01-15T10:35:00Z"
    }
  ]
}
```

<div id="ttl-configuration">
  ### TTL 配置
</div>

会话有两个 TTL 参数：

| 参数            | 默认值          | 描述                        |
| ------------- | ------------ | ------------------------- |
| `ttl`         | 600s (10 分钟) | 会话的最长存续时间 (30-3600s)      |
| `activityTtl` | 300s (5 分钟)  | 会话空闲达到该时长后自动关闭 (10-3600s) |

<div id="close-a-session">
  ### 结束会话
</div>

<CodeGroup>
  ```js Node theme={null}
  import { Firecrawl } from 'firecrawl';

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });

  await firecrawl.deleteBrowser("YOUR_SESSION_ID");
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl

  app = Firecrawl(api_key="fc-YOUR-API-KEY")

  app.delete_browser("YOUR_SESSION_ID")
  ```

  ```bash CLI theme={null}
  # 关闭当前会话
  firecrawl browser close

  # 关闭指定会话
  firecrawl browser close --session <id>
  ```

  ```bash cURL theme={null}
  curl -X DELETE "https://api.firecrawl.dev/v2/interact/YOUR_SESSION_ID" \
    -H "Authorization: Bearer $FIRECRAWL_API_KEY"
  ```
</CodeGroup>

<div id="live-view">
  ## 实时视图
</div>

每个会话的响应中都会包含一个 `liveViewUrl`，你可以将其嵌入以实时查看浏览器行为。适用于调试、演示或构建由浏览器驱动的 UI。

```json Response theme={null}
{
  "success": true,
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "cdpUrl": "wss://browser.firecrawl.dev/cdp/550e8400...?token=abc123...",
  "liveViewUrl": "https://liveview.firecrawl.dev/...",
  "interactiveLiveViewUrl": "https://liveview.firecrawl.dev/...",
  "expiresAt": "2025-01-15T10:40:00Z"
}
```

```html theme={null}
<iframe src="LIVE_VIEW_URL" width="100%" height="600" />
```

<div id="interactive-live-view">
  ### 交互式实时视图
</div>

响应中还包含一个 `interactiveLiveViewUrl`。与仅支持查看的标准实时视图不同，交互式实时视图允许用户通过嵌入的流直接点击、输入并与浏览器会话进行交互。这对于构建面向用户的浏览器界面、协同调试，或任何需要查看者直接控制浏览器的场景都非常有用。

```html theme={null}
<iframe src="INTERACTIVE_LIVE_VIEW_URL" width="100%" height="600" />
```

<div id="connecting-via-cdp">
  ## 通过 CDP 连接
</div>

每个会话都会暴露一个 CDP WebSocket URL。execute API 和 `--bash` 参数覆盖了大多数用例，但如果你需要完全的本地控制，可以直接进行连接。

<CodeGroup>
  ```typescript TypeScript theme={null}
  import { Firecrawl } from 'firecrawl';
  import { chromium } from "playwright-core";

  const firecrawl = new Firecrawl({ apiKey: "fc-YOUR-API-KEY" });
  const session = await firecrawl.browser();

  const browser = await chromium.connectOverCDP(session.cdpUrl);
  const context = browser.contexts()[0];
  const page = context.pages()[0] || (await context.newPage());

  await page.goto("https://example.com");
  console.log(await page.title());

  await browser.close();
  await firecrawl.deleteBrowser(session.id);
  ```

  ```python Python theme={null}
  from firecrawl import Firecrawl
  from playwright.sync_api import sync_playwright

  app = Firecrawl(api_key="fc-YOUR-API-KEY")
  session = app.browser()

  with sync_playwright() as p:
      browser = p.chromium.connect_over_cdp(session.cdp_url)
      context = browser.contexts[0]
      page = context.pages[0] if context.pages else context.new_page()

      page.goto("https://example.com")
      print(page.title())

      browser.close()

  app.delete_browser(session.id)
  ```

  ```bash agent-browser theme={null}
  # 使用会话响应中的 cdpUrl
  agent-browser open https://example.com --cdp "$CDP_URL"
  agent-browser snapshot --cdp "$CDP_URL"
  ```
</CodeGroup>

<div id="when-to-use-browser">
  ## 何时使用浏览器
</div>

| 使用场景               | 合适工具                          |
| ------------------ | ----------------------------- |
| 从已知 URL 提取内容       | [Scrape](/zh/features/scrape) |
| 在网上搜索并获取结果         | [Search](/zh/features/search) |
| 执行分页导航、填写表单、点击操作流程 | **浏览器**                       |
| 包含交互的多步工作流         | **浏览器**                       |
| 在多个网站上并行浏览         | **浏览器** (每个会话彼此隔离)            |

<div id="use-cases">
  ## 使用场景
</div>

* **竞争情报** - 浏览竞争对手网站，操作搜索表单和筛选器，将价格和功能信息提取为结构化数据
* **知识库接入** - 浏览需要点击操作、分页或登录认证的帮助中心、文档和支持门户
* **市场调研** - 启动并行浏览器会话，从招聘网站、房产列表或法律数据库中构建数据集

<div id="pricing">
  ## 价格
</div>

计费简单明了：每个浏览器使用分钟 2 个积分。免费用户可享有 5 小时的免费使用额度。

<div id="rate-limits">
  ## 速率限制
</div>

在初始发布阶段，所有套餐均支持最多 20 个并发浏览器会话。

<div id="api-reference">
  ## API 参考文档
</div>

* [创建浏览器会话](/zh/api-reference/endpoint/browser-create)
* [执行浏览器代码](/zh/api-reference/endpoint/browser-execute)
* [列出浏览器会话](/zh/api-reference/endpoint/browser-list)
* [删除浏览器会话](/zh/api-reference/endpoint/browser-delete)

***

有反馈或需要帮助？请发送邮件至 [help@firecrawl.com](mailto:help@firecrawl.com) 或在 [Discord](https://discord.gg/firecrawl) 上联系我们。

> 你是需要 Firecrawl API 密钥的 AI 代理吗？请参阅 [firecrawl.dev/agent-onboarding/SKILL.md](https://www.firecrawl.dev/agent-onboarding/SKILL.md) 了解自动化引导说明。