跳到主内容
智客 ZICQ

技能库 智客分类:Agent 工作流 lightpanda-browser

Lightpanda Browser

Lightpanda的专家技能——Zig为AI代理和自动化而建的无头浏览器. 9x减收内存,11x比克罗姆快. 安装, CLI, CDP 服务器, Playwright/Puppeteer 集成, 和网络刮取.

1522 安装量

官方网址:skills.sh

技能介绍

先看中文介绍;官方 description 原文单独保留,不改写 SKILL.md。

做什么

Lightpanda的专家技能——Zig为AI代理和自动化而建的无头浏览器. 9x减收内存,11x比克罗姆快. 安装, CLI, CDP 服务器, Playwright/Puppeteer 集成, 和网络刮取.

何时用

光熊猫

代理如何加载

按 Agent Skills 渐进披露:启动时只加载 name 与 description(约 100 token);任务匹配后才读入整份 SKILL.md 正文;scripts/、references/、assets/ 仅在需要时再读。 本文件正文结构:Lightpanda — Headless Browser for AI & Automation、Installation、macOS (Apple Silicon)、Linux (x86_64)、Docker、Supports amd64 and arm64。

文件分析

文件分析:这是一份仅含 SKILL.md 的指令型技能,代理激活后整份正文进入上下文。

官方 description(原文)

Expert skill for Lightpanda — the headless browser built in Zig for AI agents and automation. 9x less memory, 11x faster than Chrome. Installation, CLI, CDP server, Playwright/Puppeteer integration, and web scraping.

Lightpanda — Headless Browser for AI & AutomationInstallationmacOS (Apple Silicon)Linux (x86_64)DockerSupports amd64 and arm64CLI UsageFetch a URL (dump rendered HTML)Start CDP ServerCLI FlagsPlaywright IntegrationPuppeteer Integration

来源分类:skills.sh agent-skill

SKILL.md 与 Agent 调用

官方规范 ↗
name
lightpanda-browser
description
Expert skill for Lightpanda — the headless browser built in Zig for AI agents and automation. 9x less memory, 11x faster than Chrome. Installation, CLI, CDP server, Playwright/Puppeteer integration, and web scraping.
  1. 发现技能客户端向 Agent 提供名称与描述目录。
  2. 匹配与调用用户指定或任务匹配后,载入 SKILL.md 指令。
  3. 按需加载按步骤读取参考文档、使用脚本与素材。

具体调用语法与可用工具以目标 Agent 客户端为准。 查看调用机制说明 ↗

安装这个技能

Skills CLI ↗

先选择目标 Agent 和安装范围,保留技能包的附属文件,安装后检查客户端能否发现该技能。

交给 Agent 安装

复制安装指令给支持 Agent Skills 的代理,确认其中的目标目录与客户端匹配。

把 Agent Skill「lightpanda-browser」安装到我的项目:SKILL.md 原文与官方 description 见 https://zicq.com/zh/skills/skl-c07aaa6762fe4574-Lightpanda-%E6%B5%8F%E8%A7%88%E5%99%A8.html
请存为 .cursor/skills/lightpanda-browser/SKILL.md 或 .claude/skills/lightpanda-browser/SKILL.md,frontmatter 的 name 与 description 保持原样,不要改写。

GitHub 完整包 ↗

终端安装 · Skills CLI

需要 Node.js 与 npx。先查看仓库技能列表,确认实际名称。

npx skills add 'https://github.com/reason-machines/trending-skills' --list

npx skills add 'https://github.com/reason-machines/trending-skills' --skill 'lightpanda-browser'

CLI 会交互选择目标 Agent,默认安装到项目;用户级安装使用 -g。先通过查看命令核对仓库内容,再用 npx skills list 检查已安装技能。

阅读排版

name: lightpanda-browser description: Expert skill for Lightpanda — the headless browser built in Zig for AI agents and automation. 9x less memory, 11x faster than Chrome. Installation, CLI, CDP server, Playwright/Puppeteer integration, and web scraping. triggers:

  • lightpanda
  • headless browser for AI
  • lightweight headless browser
  • CDP automation
  • web scraping automation
  • zig browser
  • lightpanda browser
  • fast headless browser

Lightpanda — Headless Browser for AI & Automation

Skill by ara.so — Daily 2026 Skills collection

Lightpanda is a headless browser built from scratch in Zig, designed for AI agents, web scraping, and automation. It uses 9x less memory and runs 11x faster than Chrome headless.

Key facts:

  • Not based on Chromium, Blink, or WebKit — clean-slate Zig implementation
  • JavaScript execution via V8 engine
  • CDP (Chrome DevTools Protocol) compatible — works with Playwright, Puppeteer, chromedp
  • Respects robots.txt via --obey_robots flag
  • Beta status, actively developed
  • License: AGPL-3.0

Installation

macOS (Apple Silicon)

curl -L -o lightpanda https://github.com/lightpanda-io/browser/releases/download/nightly/lightpanda-aarch64-macos
chmod a+x ./lightpanda

Linux (x86_64)

curl -L -o lightpanda https://github.com/lightpanda-io/browser/releases/download/nightly/lightpanda-x86_64-linux
chmod a+x ./lightpanda

Docker

# Supports amd64 and arm64
docker run -d --name lightpanda -p 9222:9222 lightpanda/browser:nightly

CLI Usage

Fetch a URL (dump rendered HTML)

./lightpanda fetch --obey_robots --log_format pretty --log_level info https://example.com

Start CDP Server

./lightpanda serve --obey_robots --log_format pretty --log_level info --host 127.0.0.1 --port 9222

This launches a WebSocket-based CDP server for programmatic control.

CLI Flags

| Flag | Description | |------|-------------| | --obey_robots | Respect robots.txt rules | | --log_format pretty | Human-readable log output | | --log_level info | Log verbosity: debug, info, warn, error | | --host 127.0.0.1 | Bind address for CDP server | | --port 9222 | Port for CDP server | | --insecure_disable_tls_host_verification | Disable TLS verification (testing only) |

Playwright Integration

Start the CDP server, then connect Playwright to it:

import { chromium } from 'playwright-core';

const browser = await chromium.connectOverCDP('http://127.0.0.1:9222');
const context = await browser.contexts()[0] || await browser.newContext();
const page = await context.newPage();

await page.goto('https://example.com', { waitUntil: 'networkidle' });
const title = await page.title();
const content = await page.content();

console.log(`Title: ${title}`);
console.log(`HTML length: ${content.length}`);

await browser.close();

Puppeteer Integration

import puppeteer from 'puppeteer-core';

const browser = await puppeteer.connect({
  browserWSEndpoint: 'ws://127.0.0.1:9222',
});

const context = await browser.createBrowserContext();
const page = await context.newPage();

await page.goto('https://example.com', { waitUntil: 'networkidle0' });

const title = await page.title();
const text = await page.evaluate(() => document.body.innerText);

console.log(`Title: ${title}`);
console.log(`Body text: ${text.substring(0, 200)}`);

await page.close();
await browser.close();

Go (chromedp) Integration

package main

import (
    "context"
    "fmt"
    "log"

    "github.com/chromedp/chromedp"
)

func main() {
    allocCtx, cancel := chromedp.NewRemoteAllocator(context.Background(), "ws://127.0.0.1:9222")
    defer cancel()

    ctx, cancel := chromedp.NewContext(allocCtx)
    defer cancel()

    var title string
    err := chromedp.Run(ctx,
        chromedp.Navigate("https://example.com"),
        chromedp.Title(&title),
    )
    if err != nil {
        log.Fatal(err)
    }
    fmt.Println("Title:", title)
}

Python Integration

import asyncio
from playwright.async_api import async_playwright

async def main():
    async with async_playwright() as p:
        browser = await p.chromium.connect_over_cdp("http://127.0.0.1:9222")
        context = browser.contexts[0] if browser.contexts else await browser.new_context()
        page = await context.new_page()

        await page.goto("https://example.com", wait_until="networkidle")
        title = await page.title()
        content = await page.content()

        print(f"Title: {title}")
        print(f"HTML length: {len(content)}")

        await browser.close()

asyncio.run(main())

Web Scraping Patterns

Batch Page Fetching

import { chromium } from 'playwright-core';

const browser = await chromium.connectOverCDP('http://127.0.0.1:9222');
const context = await browser.newContext();

const urls = [
  'https://example.com/page1',
  'https://example.com/page2',
  'https://example.com/page3',
];

for (const url of urls) {
  const page = await context.newPage();
  await page.goto(url, { waitUntil: 'networkidle' });

  const data = await page.evaluate(() => ({
    title: document.title,
    text: document.body.innerText,
    links: [...document.querySelectorAll('a[href]')].map(a => a.href),
  }));

  console.log(JSON.stringify(data, null, 2));
  await page.close();
}

await browser.close();

Extract Structured Data

const data = await page.evaluate(() => {
  const items = document.querySelectorAll('.product-card');
  return [...items].map(item => ({
    name: item.querySelector('h2')?.textContent?.trim(),
    price: item.querySelector('.price')?.textContent?.trim(),
    link: item.querySelector('a')?.href,
  }));
});

Docker Compose (with your app)

services:
  lightpanda:
    image: lightpanda/browser:nightly
    ports:
      - "9222:9222"
    restart: unless-stopped

  scraper:
    build: .
    depends_on:
      - lightpanda
    environment:
      - BROWSER_WS_ENDPOINT=ws://lightpanda:9222

Supported Web APIs

Lightpanda supports (partial, expanding):

  • DOM tree manipulation and querying
  • JavaScript execution (V8)
  • XMLHttpRequest (XHR)
  • Fetch API
  • Cookie management
  • Network interception
  • Proxy support

Configuration

| Environment Variable | Description | |---------------------|-------------| | LIGHTPANDA_DISABLE_TELEMETRY | Set to true to opt out of usage metrics |

Performance Comparison

| Metric | Lightpanda | Chrome Headless | |--------|-----------|-----------------| | Memory | ~9x less | Baseline | | Speed | ~11x faster | Baseline | | Binary size | Small (Zig) | Large (Chromium) | | Rendering | No visual rendering | Full rendering engine |

When to Use Lightpanda

Use Lightpanda when:

  • Running AI agents that need to browse the web
  • Batch scraping at scale (memory/CPU savings matter)
  • Automating form submissions and data extraction
  • Running in containers where resources are limited
  • You need CDP compatibility but not full visual rendering

Use Chrome/Playwright instead when:

  • You need pixel-perfect screenshots or PDF generation
  • You need full Web API coverage (Lightpanda is still partial)
  • Visual regression testing
  • Testing browser-specific rendering behavior

Building from Source

Requires: Zig 0.15.2, Rust, CMake, system dependencies.

# Ubuntu/Debian dependencies
sudo apt install xz-utils ca-certificates pkg-config libglib2.0-dev clang make curl

# Build
git clone https://github.com/lightpanda-io/browser.git
cd browser
zig build

# Optional: pre-build V8 snapshot for faster startup
zig build snapshot_creator -- src/snapshot.bin
zig build -Dsnapshot_path=../../snapshot.bin

Troubleshooting

Connection refused on port 9222:

  • Ensure ./lightpanda serve is running
  • Check --host 0.0.0.0 if connecting from Docker/remote

Playwright script breaks after update:

  • Lightpanda is beta — Playwright's capability detection may behave differently across versions
  • Pin your Lightpanda version or use nightly consistently

Missing Web API support:

Links

  • GitHub: https://github.com/lightpanda-io/browser
  • Runtime APIs: https://github.com/lightpanda-io/zig-js-runtime

相关技能

Agent 工作流

Skill Creator

创造有效技能指南。 当用户想创造出新的技能(或更新现有的技能),以专业知识,工作流程,或工具集成来扩展克洛德的能力时,应该使用这种技能.

Agent 工作流

Clawdhub

使用ClawdHub CLI搜索,安装,更新并发布从taladhub.com的代理技能. 需要获取苍蝇上的新技能时使用,将安装的技能同步到最新版本或特定版本,或者发布 npm-instainddhub CLI 的新/更新的技能文件夹.

Agent 工作流

Agent Team Orchestration

管弦乐团多代理团队,任务设定周期,交接协议,审查工作流程. 使用时间: (1)建立2+特派员队伍,具有不同专业,(2)确定任务路线和生命周期(收录框_ spec_建设_审查_完成),(3)在特派员之间制定交接协议,(4)建立审查和质量关口,(5)管理特派员之间的交流和文物共享.

Agent 工作流

Superpowers

Spec-first,TDD,子代理驱动的软件开发工作流程. 当:(1)构建任何新功能或应用——触发脑暴_计划_子代理执行回路,(2)调试出一个bug或测试失败——触发系统性的根起过程,(3)用户说"让我们构建","帮助我计划","我想添加X",或"这个被打破",(4)完成一个功…