name: browser-ladder version: 1.0.0 description: Climb the browser ladder — start free, escalate only when needed. L1 (fetch) → L2 (local Playwright) → L3 (BrowserCat) → L4 (Browserless.io for CAPTCHA/bot bypass). metadata: clawdbot: emoji: "🪜" requires: bins: - node - docker env: - name: BROWSERCAT_API_KEY description: BrowserCat API key (free tier) - get at https://browsercat.com required: false - name: BROWSERLESS_TOKEN description: Browserless.io token ($10/mo) - get at https://browserless.io required: false

Browser Ladder 🪜

Climb from free to paid only when you need to.

Quick Setup

Run the setup script after installation:

./skills/browser-ladder/scripts/setup.sh

Or manually add to your .env:

# Optional - only needed for Rungs 3-4
BROWSERCAT_API_KEY=your-key    # Free: https://browsercat.com
BROWSERLESS_TOKEN=your-token   # Paid: https://browserless.io

The Ladder

┌─────────────────────────────────────────────┐
│  🪜 Rung 4: Browserless.io (Cloud Paid)     │
│  • CAPTCHA solving, bot detection bypass    │
│  • Cost: $10+/mo                            │
│  • Requires: BROWSERLESS_TOKEN              │
├─────────────────────────────────────────────┤
│  🪜 Rung 3: BrowserCat (Cloud Free)         │
│  • When local Docker fails                  │
│  • Cost: FREE (limited)                     │
│  • Requires: BROWSERCAT_API_KEY             │
├─────────────────────────────────────────────┤
│  🪜 Rung 2: Playwright Docker (Local)       │
│  • JavaScript rendering, screenshots        │
│  • Cost: FREE (CPU only)                    │
│  • Requires: Docker installed               │
├─────────────────────────────────────────────┤
│  🪜 Rung 1: web_fetch (No browser)          │
│  • Static pages, APIs, simple HTML          │
│  • Cost: FREE                               │
│  • Requires: Nothing                        │
└─────────────────────────────────────────────┘

Start at the bottom. Climb only when needed.
7w4.net小蔥技能站，你的AI助手技能庫。

When to Climb

Decision Flow

Need to access a URL
         │
         ▼
    Static content? ──YES──▶ Rung 1 (web_fetch)
         │ NO
         ▼
    JS rendering only? ──YES──▶ Rung 2 (Playwright Docker)
         │ NO                        │
         │                     Success? ──NO──▶ Rung 3
         ▼                           │ YES
    CAPTCHA/bot detection? ────────────────────▶ DONE
         │ YES
         ▼
    Rung 4 (Browserless.io) ──▶ DONE

Usage Examples

Rung 1: Static content

// Built into Clawdbot
const content = await web_fetch("https://example.com");

Rung 2: JS-rendered page

docker run --rm -v /tmp:/output mcr.microsoft.com/playwright:v1.58.0-jammy \
  npx playwright screenshot https://spa-app.com /output/shot.png

Rung 3: Cloud browser (BrowserCat)

const { chromium } = require('playwright');
const browser = await chromium.connect('wss://api.browsercat.com/connect', {
  headers: { 'Api-Key': process.env.BROWSERCAT_API_KEY }
});

Rung 4: CAPTCHA bypass (Browserless)

const { chromium } = require('playwright');
const browser = await chromium.connectOverCDP(
  `wss://production-sfo.browserless.io?token=${process.env.BROWSERLESS_TOKEN}`
);
// CAPTCHA handled automatically

Cost Optimization

1. Start low — Always try Rung 1 first
2. Cache results — Don't re-fetch unnecessarily
3. Batch requests — One browser session for multiple pages
4. Check success — Only climb if lower rung fails

Get Your Keys

Both are optional — Rungs 1-2 work without any API keys.

這個 Skill 創意不錯，用階梯式設計幫助使用者從免費方案逐步升級到付費服務，能有效控制成本。文件清晰詳細，安裝配置指令碼使用方便。但目前 Level 3 功能還沒做完，是個明顯缺憾。Level 2 依賴 Docker，對環境要求較高，配置不當容易出錯。整體完成度還可以，但需要繼續完善才能真正實用。