Browsegrab

browsegrab

한국어 문서 · llms.txt

面向本地大模型的高效浏览器 Agent — Playwright + 无障碍树 + MarkGrab，原生支持 MCP。

browsegrab 是一款轻量级浏览器自动化库，专为本地大模型（8B-35B 参数）设计。它结合了 Playwright 的无障碍树与 MarkGrab 的 HTML 转 Markdown 功能，与 browser-use 等替代方案相比，每个步骤的 Token 用量减少 5-8 倍。

功能

• 高效使用 Token：约 500-1,500 个 Token/步骤（browser-use 约为 4,000-10,000）
• 优先支持本地大模型：针对 vLLM、Ollama 和兼容 OpenAI 的端点进行优化
• 原生支持 MCP：内置 MCP 服务，包含 8 个浏览器自动化工具
• 集成 MarkGrab：将 HTML 转换为简洁的 Markdown，以提取内容
• 无障碍树 + 引用系统：无需视觉模型即可获得稳定的元素引用（e1、e2……）
• 成功模式缓存：重复工作流无需调用大模型
• 5 阶段 JSON 解析器：稳健解析本地大模型输出的操作
• 依赖精简：核心部分仅依赖 playwright + httpx

安装

pip install browsegrab
playwright install chromium

安装可选功能：

pip install browsegrab[mcp]      # MCP server support
pip install browsegrab[content]  # MarkGrab content extraction
pip install browsegrab[cli]      # CLI with rich output
pip install browsegrab[all]      # Everything

快速开始

Python API

from browsegrab import BrowseSession

async with BrowseSession() as session:
    # Navigate and get accessibility tree snapshot
    await session.navigate("https://example.com")
    snap = await session.snapshot()
    print(snap.tree_text)
    # - heading "Example Domain" [level=1]
    # - link "Learn more": [ref=e1]

    # Click using ref ID
    result = await session.click("e1")
    print(result.url)  # https://www.iana.org/help/example-domains

    # Type into search box
    await session.navigate("https://en.wikipedia.org")
    snap = await session.snapshot()
    await session.type("e4", "Python programming", submit=True)

    # Extract compressed content (AX tree + markdown)
    content = await session.extract_content()

CLI

# Accessibility tree snapshot
browsegrab snapshot https://example.com

# JSON output
browsegrab snapshot https://example.com -f json

# Extract content (AX tree + markdown)
browsegrab extract https://en.wikipedia.org/wiki/Python

# Agentic browse (requires LLM endpoint)
browsegrab browse https://example.com "Find the about page"

MCP 服务

browsegrab-mcp  # Start MCP server (stdio)

Claude Desktop / Cursor / VS Code 配置：

{
  "mcpServers": {
    "browsegrab": {
      "command": "browsegrab-mcp"
    }
  }
}

8 个 MCP 工具：browser_navigate、browser_click、browser_type、browser_snapshot、browser_scroll、browser_extract_content、browser_go_back、browser_wait

工作原理

Agent 浏览循环

flowchart LR
    A["🌐 URL + Goal"] --> B["Navigate"]
    B --> C["AX Tree Snapshot\n~200–500 tokens"]
    C --> D{"LLM\nDecision"}
    D -->|"click / type / scroll"| E["Execute Action"]
    E --> C
    D -->|"goal reached"| F["Extract Content\n(MarkGrab)"]
    F --> G["✅ Result"]

Token 效率

browsegrab 将结构（无障碍树）与内容（MarkGrab Markdown）分开，只向大模型发送其所需的信息：

flowchart TD
    A["Raw HTML"] --> B["Accessibility Tree"]
    A --> C["MarkGrab Markdown"]
    B --> D["Structure: ~200–500 tokens\nInteractive elements with ref IDs"]
    C --> E["Content: ~300–800 tokens\nClean markdown · on-demand"]
    D --> F["Combined: ~500–1,300 tokens/step\n⚡ 5–8× fewer than browser-use"]
    E --> F

Token 效率（实测）

页面	交互元素	Token 数	browser-use 等效值
example.com	1	~60	~500+
Wikipedia 文章	452	~1,254	~10,000+

架构

browsegrab/
├── config.py                 # Dataclass configs (env var loading)
├── result.py                 # Result types (ActionResult, BrowseResult, ...)
├── session.py                # BrowseSession orchestrator
├── browser/
│   ├── manager.py            # Playwright lifecycle (async context manager)
│   ├── snapshot.py           # Accessibility tree + ref system
│   ├── selectors.py          # 4-strategy selector resolver
│   └── actions.py            # navigate, click, type, scroll, go_back, wait
├── dom/
│   ├── ref_map.py            # ref ID ↔ element bidirectional mapping
│   └── compress.py           # AX tree + MarkGrab → compressed context
├── llm/
│   ├── base.py               # LLMProvider ABC
│   ├── provider.py           # vLLM, Ollama, OpenAI-compatible
│   ├── prompt.py             # System prompts (~400 tokens)
│   └── parse.py              # 5-stage JSON fallback parser
├── agent/
│   ├── history.py            # Sliding window history compression
│   ├── cache.py              # Domain-based success pattern cache
│   └── loop_guard.py         # Duplicate action detection
├── __main__.py               # CLI (click)
└── mcp_server.py             # FastMCP server (8 tools)

配置

通过环境变量配置所有设置（前缀为 BROWSEGRAB_*）：

# Browser
BROWSEGRAB_BROWSER_HEADLESS=true
BROWSEGRAB_BROWSER_TIMEOUT_MS=30000

# LLM (for agentic browse)
BROWSEGRAB_LLM_PROVIDER=vllm          # vllm | ollama | openai
BROWSEGRAB_LLM_BASE_URL=http://localhost:8000/v1
BROWSEGRAB_LLM_MODEL=Qwen/Qwen3.5-32B-AWQ

# Agent
BROWSEGRAB_AGENT_MAX_STEPS=10
BROWSEGRAB_AGENT_ENABLE_CACHE=true

QuartzUnit 生态系统的一部分

库	作用
markgrab	被动提取（URL → Markdown）
snapgrab	被动捕获（URL → 截图）
docpick	文档 OCR → 结构化 JSON
browsegrab	主动自动化（目标 → 浏览器操作 → 结果）

开发

git clone https://github.com/QuartzUnit/browsegrab.git
cd browsegrab
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
playwright install chromium

# Unit tests (no browser needed)
pytest tests/ -m "not e2e"

# Full suite including E2E
pytest tests/ -v

许可证

MIT

---

_{属于 QuartzUnit 生态系统 — 用于数据收集、提取、搜索和 AI Agent 安全的可组合 Python 库。}