AgentSkillsCN

headless-browser

浏览JavaScript密集型网站，利用OpenClaw内置浏览器工具提取文本与截图（无隐身模式）。

SKILL.md
--- frontmatter
name: headless-browser
description: Browse JS-heavy sites and extract text/snapshots using OpenClaw's built-in browser tool (no stealth).
metadata: {"openclaw":{"emoji":"🌐","requires":{"bins":["openclaw"]}}}

Headless Browser (OpenClaw Browser Tool)

Use this skill when the user asks to:

  • •open a URL and extract visible text
  • •take a screenshot
  • •navigate JS-heavy sites

Policy / constraints

  • •Do not attempt to bypass logins, paywalls, or bot protections.
  • •Do not use stealth plugins or "anti-bot evasion" techniques.
  • •If a page is blocked, propose alternatives (official APIs, RSS, public mirrors, or ask the user to provide the content).

How to use (CLI)

Start the managed browser (if needed):

bash
openclaw browser status
openclaw browser start

Open a page:

bash
openclaw browser open "https://example.com"

Extract a snapshot (prefer --format md for readable output):

bash
openclaw browser snapshot --format md --limit 200

Take a screenshot:

bash
openclaw browser screenshot --full-page

Job search guidance

  • •Prefer public job boards and company career pages.
  • •If the user mentions LinkedIn, prefer public listings and metadata; do not attempt to access restricted pages.