browser_guide

Agent Skills › siddsachar/row-bot › browser_guide

browser_guide

GitHub

提供浏览器自动化工作流指导，支持通过工具操控真实浏览器窗口进行页面导航、元素交互及标签页管理。适用于需用户辅助处理登录或验证码的复杂网页操作场景。

tool_guides/browser_guide/SKILL.md siddsachar/row-bot

Trigger Scenarios

用户请求浏览网页或打开浏览器需要点击、滚动或填写表单等页面交互执行基于快照的自动化浏览任务

Install

npx skills add siddsachar/row-bot --skill browser_guide -g -y

More Options

Non-standard path

npx skills add https://github.com/siddsachar/row-bot/tree/main/tool_guides/browser_guide -g -y

Use without installing

npx skills use siddsachar/row-bot@browser_guide

指定 Agent (Claude Code)

npx skills add siddsachar/row-bot --skill browser_guide -a claude-code -g -y

安装 repo 全部 skill

npx skills add siddsachar/row-bot --all -g -y

预览 repo 内 skill

npx skills add siddsachar/row-bot --list

SKILL.md

Frontmatter

{
    "icon": "🌐",
    "name": "browser_guide",
    "tags": [],
    "tools": [
        "browser"
    ],
    "description": "Guidance for browser automation workflows.",
    "display_name": "Browser Guide"
}

BROWSER AUTOMATION (experimental):

You have a browser tool that opens a REAL visible browser window. The user can see the browser and interact with it too (e.g. to type passwords or solve CAPTCHAs).
Workflow: browser_navigate → read the snapshot → browser_click / browser_type / browser_scroll → read updated snapshot → repeat.
You can manage tabs with browser_tab: list open tabs, switch between them, open new tabs, or close tabs by index.
Use browser_back to navigate back to the previous page.
Each snapshot lists interactive elements with numbered refs like [1] button "Submit", [2] input[text] "Search". Use the ref number to click or type.
IMPORTANT: refs become stale after any navigation or page change. Always use the refs from the MOST RECENT snapshot only.
If you encounter a login page or CAPTCHA, tell the user to handle it in the browser window, then call browser_snapshot to see the result.
When the user says 'browse', 'open in the browser', or asks you to interact with a page (click, scroll, fill forms), ALWAYS use the browser_* tools. Use read_url ONLY when you need raw text from a URL and the user has NOT mentioned the browser.