Skip to content
Back to skills

Browser Use

ASecurity

Use when an AI agent needs to control a browser, automate web tasks, scrape pages, fill forms, or click buttons autonomously. Triggers on: 'browser automation', 'web agent', 'browser-use', 'AI browse', 'tự động duyệt web', 'điều khiển trình duyệt', 'scrape with AI', 'click button automatically', 'fill form automatically', 'web task automation'.

  • 10 stars
  • 0 votes
  • 0 copies
  • 6 views
  • Added September 9, 2026
ai-agentspythongobashgitapisecurity

Works with

  • cli
  • api

Security analysis

A96/100
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro shows the line behind each finding and how to fix it

Scanned September 9, 2026

npx -y skills add yanacuti1121/Yana-AI --skill browser-use --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Browser Use?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Browser Use
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/yanacuti1121-browser-use/badge)](https://www.skillsdirectory.com/skills/yanacuti1121-browser-use)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: browser-use
description: "Use when an AI agent needs to control a browser, automate web tasks, scrape pages, fill forms, or click buttons autonomously. Triggers on: 'browser automation', 'web agent', 'browser-use', 'AI browse', 'tự động duyệt web', 'điều khiển trình duyệt', 'scrape with AI', 'click button automatically', 'fill form automatically', 'web task automation'."
---

# Browser-Use Skill

Cho AI agent điều khiển trình duyệt thực sự — click, type, scroll, scrape, fill form.
Source: [browser-use/browser-use](https://github.com/browser-use/browser-use) (95K⭐, MIT)

## Install

```bash
pip install browser-use
uvx browser-use install   # cài Chromium nếu chưa có
```

## Workflow

### Step 1 — Setup env

```bash
# .env
ANTHROPIC_API_KEY=your-key
# hoặc OPENAI_API_KEY, GOOGLE_API_KEY
```

### Step 2 — Basic agent

```python
import asyncio
from browser_use import Agent, Browser
from langchain_anthropic import ChatAnthropic

async def main():
    browser = Browser()
    agent = Agent(
        task="Tìm giá iPhone 15 Pro trên tgdd.vn và trả về kết quả",
        llm=ChatAnthropic(model="claude-sonnet-4-6"),
        browser=browser,
    )
    result = await agent.run()
    print(result.final_result())
    await browser.close()

asyncio.run(main())
```

### Step 3 — Với context tùy chỉnh (login session, cookies)

```python
from browser_use import Agent, Browser, BrowserConfig, BrowserContextConfig

browser = Browser(
    config=BrowserConfig(
        headless=True,           # chạy ẩn
        disable_security=False,
    )
)

async with await browser.new_context(
    config=BrowserContextConfig(
        cookies=[{"name": "session", "value": "abc", "domain": ".example.com"}]
    )
) as ctx:
    agent = Agent(task="...", llm=llm, browser_context=ctx)
    await agent.run()
```

### Step 4 — Multi-step task với custom actions

```python
from browser_use import Agent, Controller
from browser_use.browser.context import BrowserContext

controller = Controller()

@controller.action("Lưu dữ liệu vào file", param_model=SaveParams)
async def save_data(params: SaveParams, browser: BrowserContext):
    page = await browser.get_current_page()
    content = await page.content()
    with open(params.filename, "w") as f:
        f.write(content)
    return ActionResult(extracted_content="Saved", include_in_memory=True)

agent = Agent(task="...", llm=llm, controller=controller)
await agent.run(max_steps=20)
```

### Step 5 — Khai thác kết quả

```python
history = await agent.run()

# Kết quả cuối
print(history.final_result())

# Toàn bộ actions đã thực hiện
for action in history.model_actions():
    print(action)

# Screenshots từng bước
for screenshot in history.screenshots():
    # screenshot là base64 PNG
    pass

# URLs đã visit
print(history.urls())
```

## Use cases thường gặp

```python
# Scrape + extract structured data
agent = Agent(
    task="""
    Vào trang https://example.com/products
    Lấy tên, giá, và rating của 10 sản phẩm đầu tiên
    Trả về dạng JSON list
    """,
    llm=llm,
)

# Form automation
agent = Agent(
    task="""
    Vào trang https://form.example.com
    Điền: Name='Test', Email='test@mail.com', Message='Hello'
    Submit form
    Xác nhận đã gửi thành công
    """,
    llm=llm,
)
```

## Options

| Option | Default | Mô tả |
|--------|---------|--------|
| `headless` | `True` | Chạy không có cửa sổ |
| `max_steps` | `100` | Số bước tối đa |
| `use_vision` | `True` | Dùng screenshot để hiểu trang |
| `save_conversation_path` | `None` | Lưu toàn bộ conversation log |

## Lỗi thường gặp

| Lỗi | Fix |
|-----|-----|
| `Chromium not found` | Chạy `uvx browser-use install` |
| Agent loop không dừng | Giảm `max_steps`, thêm điều kiện dừng rõ ràng vào task |
| Page không load xong | Thêm "đợi trang load xong trước khi thao tác" vào task description |

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.