/install chromeopencrawl
OpenCrawl Skill
Use this skill to crawl any JavaScript-rendered webpage using real Chrome browsers from a distributed worker pool. Unlike headless browser solutions (Puppeteer/Playwright), OpenCrawl requires zero local browser installation — ideal for VPS and cloud environments.
Quick Start (use our public server)
- Visit http://39.105.206.76:9877 and click "Register" to get a free API Key (100 credits included)
- Set environment variables:
OPENCRAWL_API_KEY=ak_your_key_here OPENCRAWL_API_URL=http://39.105.206.76:9877 - Start crawling!
Self-hosted (deploy your own server)
If you prefer to run your own OpenCrawl server, see the full deployment guide: https://github.com/hlyylly/OpenCrawl
Then set OPENCRAWL_API_URL to your own server address.
How it works: Your request → OpenCrawl server → dispatched to a real Chrome browser worker → page rendered with full JavaScript → content extracted → uploaded to Cloudflare R2 → download URL returned to you.
Errors: On failure the script writes a JSON error to stderr and exits with code 1.
Tools
1. Crawl Page
Use this to get the full rendered text content of any webpage, including JavaScript-rendered content that simple HTTP requests cannot retrieve.
Command:
python3 {baseDir}/tools/crawl.py --url "https://example.com"
Examples:
# Crawl a full page
python3 {baseDir}/tools/crawl.py --url "https://www.smzdm.com/p/170177008/"
# Crawl with CSS selector to extract specific content
python3 {baseDir}/tools/crawl.py --url "https://example.com" --selector ".article-content"
# Output raw JSON response (includes downloadUrl)
python3 {baseDir}/tools/crawl.py --url "https://example.com" --raw
Optional flags:
--selector ".css-selector"— Extract only matching elements--mode lite— Lite mode: no images/CSS, faster, 0.1 credit (default: full)--raw— Output full JSON response instead of just the text content--timeout 60— Custom timeout in seconds (default: 60)
2. Search (Brave Search API Compatible)
Use this to search the web using multiple search engines (DuckDuckGo + Google + Bing + Baidu) through real Chrome browsers. Returns structured results compatible with Brave Search API format.
Command:
python3 {baseDir}/tools/crawl.py --search "your search query"
Examples:
# Lite search — DuckDuckGo only (0.1 credit)
python3 {baseDir}/tools/crawl.py --search "python web scraping"
# Full search — 4 engines parallel (3 credits, 20-30 deduplicated results)
python3 {baseDir}/tools/crawl.py --search "python web scraping" --mode full
4. Check Balance
Use this to check how many credits remain on the API key.
Command:
python3 {baseDir}/tools/crawl.py --balance
5. Check Status
Use this to check the OpenCrawl platform status — how many workers are online, tasks completed, etc.
Command:
python3 {baseDir}/tools/crawl.py --status
Summary
| Action | Argument | Example |
|---|---|---|
| Crawl (full) | --url |
python3 {baseDir}/tools/crawl.py --url "https://example.com" |
| Crawl (lite) | --url --mode lite |
python3 {baseDir}/tools/crawl.py --url "https://example.com" --mode lite |
| Search (lite) | --search |
python3 {baseDir}/tools/crawl.py --search "python tutorial" |
| Search (full) | --search --mode full |
python3 {baseDir}/tools/crawl.py --search "python tutorial" --mode full |
| Check balance | --balance |
python3 {baseDir}/tools/crawl.py --balance |
| Check status | --status |
python3 {baseDir}/tools/crawl.py --status |
Output: Crawl → rendered page text (or JSON with --raw). Search → JSON with web.results[] (Brave compatible). Balance → JSON. Status → JSON.
Requirements: Python 3.8+, requests library. No browser installation needed.
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install chromeopencrawl - 安装完成后,直接呼叫该 Skill 的名称或使用
/chromeopencrawl触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
crawl 是什么?
Crawl any JavaScript-rendered webpage through distributed real Chrome browsers. No local browser needed — perfect for headless VPS. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 227 次。
如何安装 crawl?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install chromeopencrawl」即可一键安装,无需额外配置。
crawl 是免费的吗?
是的,crawl 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
crawl 支持哪些平台?
crawl 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 crawl?
由 hlyylly(@hlyylly)开发并维护,当前版本 v1.0.2。