DrissionPageMCP
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@DrissionPageMCP Open https://example.com and take a screenshot."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
DrissionPage MCP Server -- 骚神出品
基于DrissionPage和FastMCP的浏览器自动化MCP服务器,提供丰富的浏览器操作API供AI调用。
项目简介

DrissionPage MCP 是一个基于 DrissionPage 和 FastMCP 的浏览器自动化MCP server服务器,它提供了一系列强大的浏览器操作 API,让您能够轻松通过AI实现网页自动化操作。
主要特性
支持浏览器的打开、关闭和连接管理
提供丰富的页面元素操作方法
支持 JavaScript 代码执行
支持 CDP 协议操作
提供便捷的文件下载功能
支持键盘按键模拟
支持页面截图功能
增加 网页后台监听数据包的功能
增加自动上传下载文件功能
Python要求
Python >= 3.9
pip(最新版本)
uv (最新版本)
浏览器要求
Chrome 浏览器(推荐 90 及以上版本)
必需的Python包
drissionpage >= 4.1.0.18
fastmcp >= 2.4.0
uv
Related MCP server: DrissionPage MCP Server
安装说明
把本仓库git clone到本地,核心文件是main.py:
首先要进行💖MCP安装环境准备工作
安装到Cursor编辑器

安装到vscode编辑器

请将以下配置代码粘贴到编辑器的mcpServers设置中(请填写你自己电脑上 main.py 文件的绝对路径):
{
"mcpServers": {
"DrissionPageMCP": {
"type": "stdio",
"command": "uv",
"args": ["--directory", "D:\\test10\\DrissionPageMCP", "run", "main.py"]
}
}
}新增mcp配置 ,填写下面的配置:
"DrissionPageMCP": {
"type": "stdio",
"command": "uv",
"args": ["--directory", "D:\\test10\\DrissionPageMCP", "run", "main.py"]
} 注意事项:
请根据实际路径修改
args中的路径Windows中路径中的反斜杠需要转义(使用
\\)确保
uv命令在系统PATH中可用
调试命令
调试
npx -y @modelcontextprotocol/inspector uv run D:\\test10\\DrissionPageMCP\\main.py或者
mcp dev D:\\test10\\DrissionPageMCP\\main.py更新日志
v0.1.3
增加 自动上传下载文件功能
v0.1.2
增加 网页后台监听数据包的功能
v0.1.0
初始版本发布
实现基本的浏览器控制功能
提供元素操作 API
Available Tools
29 toolsclick_by_containing_textA
根据包含指定文本的方式点击网页元素。
参数: content: 要查找的文本内容。 index: 当匹配到多个元素时指定要点击的索引,默认不指定。
返回: 点击结果说明,或错误提示。
Inputs: content (required), index.
click_by_xpathA
通过xpath点击当前标签页中某个元素,最好先获取页面dom信息,再决定Xpath的写法
Inputs: xpath (required).
connect_or_open_browserB
用DrissionPage 打开或接管已打开的浏览器,参数通过字典传递。 必要参数: config (dict): 可选键包括 、debug_port、browser_path、headless 返回: dict: 浏览器信息
Inputs: config.
download_fileC
控制浏览器下载文件到指定路径
Args: url (str): 文件的URL地址 path (str): 文件保存的路径 rename (str): 重命名文件名
Returns: str: 下载结果信息
Inputs: url (required), path (required), rename (required).
dragA
将元素拖动到指定偏移位置
Args: xpath: 要拖动的元素xpath路径 offset_x: x轴偏移量(像素) offset_y: y轴偏移量(像素) duration: 拖动持续时间(毫秒),默认为1000
Returns: dict: 包含偏移量和持续时间的字典,格式为{"offset_x": int, "offset_y": int, "duration": int} 或 str: 当元素不存在时返回错误信息
Raises: 无显式抛出异常,但内部可能因元素不存在而返回错误信息
Inputs: xpath (required), offset_x (required), offset_y (required), duration.
getA
在当前标签页打开一个网址
Inputs: url (required).
get_body_textB
获取当前标签页的body的文本内容
Inputs: no named parameters.
get_cdp_event_dataB
获取CDP事件回调函数收集到的数据
Inputs: no named parameters.
get_current_tab_infoA
获取当前标签页的信息,包括url, title, id
Inputs: no named parameters.
get_current_tab_screenshotA
获取当前标签页的网页截图
Returns: bytes: 截图的二进制数据
Inputs: no named parameters.
get_current_tab_screenshot_as_fileB
获取当前标签页的屏幕截图并保存为文件
Args: path (str): 截图保存路径,默认为当前目录
Returns: str: 截图的文件路径
Inputs: name, path.
get_DrissionPage_code_guideC
获取 DrissionPage 代码指南
Inputs: no named parameters.
getInputElementsInfoA
获取当前标签页的所有可进行输入操作的元素,对元素进行输入操作前优先使用这个方法
Inputs: no named parameters.
get_response_listener_dataB
获取监听到的数据,返回数据列表
Inputs: no named parameters.
getSimplifiedDomTreeD
Inputs: no named parameters.
get_url_with_response_listenerB
开启一个新的标签页,设置监听,访问tab_url, tab_url: 被监听的标签页的url mimeType: 需要监听的接收的数据包的mimeType类型 url_include: 需要监听的接收的数据包的url包含的关键字 refresh: 是否刷新页面,
Inputs: tab_url (required), mimeType (required), url_include.
get_versionC
获取版本号
Inputs: no named parameters.
input_by_xapthA
通过xpath给当前标签页中某个元素输入内容,最好先判断元素是否存在
Args: xpath (str): 元素的XPath表达式 input_value (str): 要输入的内容 clear_first (bool): 是否先清除已有内容,默认为True
Returns: Any: 输入操作的结果,如果元素不存在则返回错误信息
Inputs: xpath (required), input_value (required), clear_first.
listen_cdp_eventA
设置监听CDP事件
应该先运行cdp 命令 激活对应的域,比如 Network.enable
Inputs: event_name (required).
move_toA
鼠标移动悬停到指定xpath的元素上
Inputs: xpath (required).
new_tabC
用DrissionPage 控制的浏览器,打开新标签页并 打开一个网址
Inputs: url (required).
response_listener_stopC
关闭监听网页发送的数据包
Inputs: clear_data.
run_cdpB
在当前标签页中运行谷歌CDP协议代码并获取结果
Args:
cmd: CDP协议命令
**cmd_args: CDP命令参数Returns: Any: CDP命令执行结果
Note: 举例1说明 run_cdp('Page.stopLoading') 举例2说明 run_cdp('Page.navigate', url='https://example.com')
Inputs: cmd (required), cmd_args (required).
run_jsA
在当前标签页中运行JavaScript代码并返回执行结果 查找网页元素,获取元素信息,操作网页元素优先使用这个方法
Args: js_code (str): 要执行的JavaScript代码
Returns: Any: JavaScript代码执行结果
Note:
想要获取执行的js代码的返回值,可以在js_code中使用return语句。
想要获取异步函数的返回值,可以参考下面代码
return (async (url) => {
const response = await fetch(url);
const data = await response.json();
return data;
})("https://www.baidu.com/");
Inputs: js_code (required).
save_dict_to_sqliteA
将字典或JSON字符串保存到SQLite数据库中。
参数: data (dict or list of dict or str): 字典、列表字典,或JSON字符串。 db_path (str): SQLite 数据库文件路径。 table_name (str): 要创建的表名。
Inputs: data (required), db_path, table_name.
send_enterA
向当前页面发送 enter 回车键
Inputs: no named parameters.
send_keyC
向当前标签页发送特殊按键
Inputs: key (required).
upload_fileA
点击当网页上的 元素触发上传文件的操作,上传file_path文件到当前网页
Args:
file_path (str): 要上传的文件路径
Returns: str: 上传结果信息,如果元素不存在则返回错误信息
Inputs: file_path (required).
waitD
等待a秒
Inputs: a (required).
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 29 tools
Most tools target distinct actions like navigation, clicking, input, screenshots, and network listening. A few pairs (send_enter/send_key, screenshot/screenshot_as_file) are closely related, but descriptions make them distinguishable.
The naming is predominantly snake_case verb phrases, but getInputElementsInfo and getSimplifiedDomTree break the convention, and input_by_xapth contains a typo. Overall it remains readable and mostly predictable.
With 29 tools, the server feels over-scoped, especially with redundant screenshot/listener functions and an unrelated save_dict_to_sqlite. The count exceeds the comfortable range and could be reduced by consolidating similar tools.
Core browser automation workflows are covered: navigation, tabs, element interaction, JS/CDP, file transfer, screenshots, and network listening. Gaps like closing tabs or cookie management are workaroundable via run_js or run_cdp.
Related MCP Connectors
Web scraping, browser automation, screenshots and AI data extraction for AI agents via MCP.
Control a real browser from AI agents: navigate, click, type, screenshot, read the page.
Stealth web browser for agents: search, fetch, click, download and type in persistent MCP sessions.
Web search, browser automation, scraping, crawling and CAPTCHA solving for AI agents.
Related MCP Servers
- AlicenseBqualityDmaintenanceAn MCP server providing robust browser automation tools for AI assistants, including page navigation, element interaction, and screenshot capabilities. It leverages the DrissionPage library to enable standardized DOM analysis, network monitoring, and complex web task automation.17MIT
- AlicenseCqualityBmaintenanceA professional browser automation server that enables MCP clients to perform structured web navigation, element interaction, and data extraction using the DrissionPage framework. It features 14 deterministic tools optimized for LLMs to automate web workflows efficiently without relying on vision-based models.69152 PyPI482Apache 2.0
- FlicenseNot gradedqualityDmaintenanceEnables AI assistants to control and interact with a Chrome browser via MCP, providing tools for navigation, screenshots, clicking, form filling, content extraction, and tab management.1-
- FlicenseBqualityBmaintenanceEnables MCP clients to drive browser automation for web navigation, element interaction, and multi-account isolation via DrissionPage 5.0.56-