dev-browser（开发浏览器）

Dev Browser Skill

Browser automation that maintains page state across script executions. Write small, focused scripts to accomplish tasks incrementally. Once you've proven out part of a workflow and there is repeated work to be done, you can write a script to do the repeated work in a single execution.

Choosing Your Approach

• Local/source-available sites: Read the source code first to write selectors directly
• Unknown page layouts: Use getAISnapshot() to discover elements and selectSnapshotRef() to interact with them
• Visual feedback: Take screenshots to see what the user sees

Setup

Two modes available. Ask the user if unclear which to use.

Standalone Mode (Default)

Launches a new Chromium browser for fresh automation sessions.

./skills/dev-browser/server.sh &

Options:

• --headless - Run in headless mode (no visible browser window)
• --stealth - Enable anti-detection mode (default: ON)
• --no-stealth - Disable stealth mode (use plain browser)
• --driver [playwright|patchright|auto] - Choose browser driver (default: auto)

- auto - Auto-install and use patchright, fallback to playwright with patches - patchright - Force use of patchright - playwright - Use standard playwright

• --no-flaresolverr - Disable FlareSolverr auto-start (default: enabled if Docker available)
• --stop - Stop all dev-browser services and cleanup
• --help - Show help message

Wait for the Ready message before running scripts.

By default, the server automatically:

1. Installs Patchright (if not present) for stealth automation
2. Enables stealth mode with browser args and runtime patches
3. Starts FlareSolverr Docker container for Cloudflare bypass (requires Docker)

To disable automatic features:

# Disable stealth entirely
./skills/dev-browser/server.sh --no-stealth &

# Disable FlareSolverr auto-start
./skills/dev-browser/server.sh --no-flaresolverr &

# Use plain Playwright without anti-detection
./skills/dev-browser/server.sh --no-stealth --driver playwright &

Examples:

# Default: Full anti-detection (stealth + patchright + flaresolverr)
./skills/dev-browser/server.sh &

# Headless mode with full anti-detection
./skills/dev-browser/server.sh --headless &

# Force specific driver
./skills/dev-browser/server.sh --driver patchright &

# Minimal setup without any anti-detection
./skills/dev-browser/server.sh --no-stealth --no-flaresolverr &

Extension Mode

Connects to user's existing Chrome browser. Use this when:

• The user is already logged into sites and wants you to do things behind an authed experience that isn't local dev.
• The user asks you to use the extension

Important: The core flow is still the same. You create named pages inside of their browser.

Start the relay server:

cd skills/dev-browser && npm i && npm run start-extension &

Wait for Waiting for extension to connect... followed by Extension connected in the console. To know that a client has connected and the browser is ready to be controlled. Workflow:

1. Scripts call client.page("name") just like the normal mode to create new pages / connect to existing ones.
2. Automation runs on the user's actual browser session

If the extension hasn't connected yet, tell the user to launch and activate it. Download link: https://github.com/SawyerHood/dev-browser/releases

Writing Scripts

Run all scripts from skills/dev-browser/ directory. The @/ import alias requires this directory's config.

Execute scripts inline using heredocs:

cd skills/dev-browser && npx tsx <<'EOF'
import { connect, waitForPageLoad } from "@/client.js";

const client = await connect();
// Create page with custom viewport size (optional)
const page = await client.page("example", { viewport: { width: 1920, height: 1080 } });

await page.goto("https://example.com");
await waitForPageLoad(page);

console.log({ title: await page.title(), url: page.url() });
await client.disconnect();
EOF

Write to tmp/ files only when the script needs reuse, is complex, or user explicitly requests it.

Key Principles

1. Small scripts: Each script does ONE thing (navigate, click, fill, check)
2. Evaluate state: Log/return state at the end to decide next steps
3. Descriptive page names: Use "checkout", "login", not "main"
4. Disconnect to exit: await client.disconnect() - pages persist on server
5. Plain JS in evaluate: page.evaluate() runs in browser - no TypeScript syntax

Workflow Loop

Follow this pattern for complex tasks:

1. Write a script to perform one action
2. Run it and observe the output
3. Evaluate - did it work? What's the current state?
4. Decide - is the task complete or do we need another script?
5. Repeat until task is done

No TypeScript in Browser Context

Code passed to page.evaluate() runs in the browser, which doesn't understand TypeScript:

// ✅ Correct: plain JavaScript
const text = await page.evaluate(() => {
  return document.body.innerText;
});

// ❌ Wrong: TypeScript syntax will fail at runtime
const text = await page.evaluate(() => {
  const el: HTMLElement = document.body; // Type annotation breaks in browser!
  return el.innerText;
});

Scraping Data

For scraping large datasets, intercept and replay network requests rather than scrolling the DOM. See references/scraping.md for the complete guide covering request capture, schema discovery, and paginated API replay.

Client API

const client = await connect();

// Get or create named page (viewport only applies to new pages)
const page = await client.page("name");
const pageWithSize = await client.page("name", { viewport: { width: 1920, height: 1080 } });

const pages = await client.list(); // List all page names
await client.close("name"); // Close a page
await client.disconnect(); // Disconnect (pages persist)

// ARIA Snapshot methods
const snapshot = await client.getAISnapshot("name"); // Get accessibility tree
const element = await client.selectSnapshotRef("name", "e5"); // Get element by ref

The page object is a standard Playwright Page.

Waiting

import { waitForPageLoad } from "@/client.js";

await waitForPageLoad(page); // After navigation
await page.waitForSelector(".results"); // For specific elements
await page.waitForURL("**/success"); // For specific URL

Inspecting Page State

Screenshots

await page.screenshot({ path: "tmp/screenshot.png" });
await page.screenshot({ path: "tmp/full.png", fullPage: true });

ARIA Snapshot (Element Discovery)

Use getAISnapshot() to discover page elements. Returns YAML-formatted accessibility tree:

- banner:
  - link "Hacker News" [ref=e1]
  - navigation:
    - link "new" [ref=e2]
- main:
  - list:
    - listitem:
      - link "Article Title" [ref=e8]
      - link "328 comments" [ref=e9]
- contentinfo:
  - textbox [ref=e10]
    - /placeholder: "Search"

Interpreting refs:

• [ref=eN] - Element reference for interaction (visible, clickable elements only)
• [checked], [disabled], [expanded] - Element states
• [level=N] - Heading level
• /url:, /placeholder: - Element properties

Interacting with refs:

const snapshot = await client.getAISnapshot("hackernews");
console.log(snapshot); // Find the ref you need

const element = await client.selectSnapshotRef("hackernews", "e2");
await element.click();

Error Recovery

Page state persists after failures. Debug with:

cd skills/dev-browser && npx tsx <<'EOF'
import { connect } from "@/client.js";

const client = await connect();
const page = await client.page("hackernews");

await page.screenshot({ path: "tmp/debug.png" });
console.log({
  url: page.url(),
  title: await page.title(),
  bodyText: await page.textContent("body").then((t) => t?.slice(0, 200)),
});

await client.disconnect();
EOF

Anti-Detection Features (Auto-Enabled by Default)

By default, the skill automatically enables all anti-detection features for protected sites with bot detection (Cloudflare, DataDome, etc.):

How It Works

Patchright Auto-Install: When you start the server with defaults (./server.sh):

1. Checks if patchright is installed
2. If not found → automatically installs it via npm
3. Uses Patchright with built-in stealth capabilities
4. Falls back to Playwright with manual patches if install fails

Stealth Mode (Default ON): Stealth mode is now enabled by default and includes:

1. Modified browser launch args (removes --enable-automation, etc.)
2. If using Playwright: runtime patches to hide navigator.webdriver and other indicators
3. If using Patchright: native stealth features handle evasion automatically
4. All new pages get stealth patches applied

FlareSolverr (Auto-Start): FlareSolverr Docker container is automatically started when:

1. Docker is installed and running
2. --no-flaresolverr flag was NOT used
3. A container named dev-browser-flaresolverr doesn't already exist

The container runs on port 8191 and is ready for bypassCloudflare() calls.

Stopping Services

To stop all dev-browser services gracefully:

# Stop all services (server, browser, FlareSolverr Docker)
./skills/dev-browser/server.sh --stop

This will:

1. Stop the dev-browser HTTP server (port 9222)
2. Close all browser contexts and pages
3. Stop the FlareSolverr Docker container
4. Clean up any stale Chrome processes

Disabling Anti-Detection

If you need a standard browser without anti-detection:

# Disable all anti-detection features
./skills/dev-browser/server.sh --no-stealth --no-flaresolverr --driver playwright

Prerequisites

All anti-detection features are auto-enabled by default when you run ./server.sh. No manual setup required.

Optional: Pre-install Patchright (to avoid install delay on first run):

cd skills/dev-browser && npm install patchright

Optional: Ensure Docker is running (for FlareSolverr auto-start):

# FlareSolverr requires Docker. If Docker is not running, the skill will skip it with a warning.
docker info

To completely disable anti-detection and use a standard browser:

./skills/dev-browser/server.sh --no-stealth --no-flaresolverr --driver playwright

Stealth Mode

Apply runtime anti-detection patches to a page to hide automation flags:

import { connect, applyStealthMode } from "@/client.js";

const client = await connect();
const page = await client.page("example");

// Apply stealth patches before navigating
await applyStealthMode(page);

await page.goto("https://example.com");
console.log("Page title:", await page.title());

await client.disconnect();

Cloudflare Bypass

For sites protected by Cloudflare challenge pages:

import { connect, bypassCloudflare } from "@/client.js";

const client = await connect();
const page = await client.page("protected");

// Use FlareSolverr to solve the challenge and inject cookies
await bypassCloudflare(page, "https://protected-site.com", {
  maxTimeout: 60000,
});

// Page now has valid Cloudflare session
console.log("After bypass:", await page.title());

await client.disconnect();

The bypassCloudflare function:

1. Sends the URL to FlareSolverr to solve the challenge
2. Receives authentication cookies
3. Injects cookies into the Playwright context
4. Navigates to the target URL

Using Patchright Directly

To launch a browser with Patchright directly (for full stealth capabilities):

import { launchBrowser, USER_AGENTS } from "@/client.js";

const context = await launchBrowser({
  usePatchright: true,
  stealth: true,
  headless: false,
  userAgent: USER_AGENTS.chrome.windows,
  viewport: { width: 1920, height: 1080 },
});

const page = await context.newPage();
await page.goto("https://example.com");

Advanced: FlareSolverr Client

For more control over challenge solving, use the FlareSolverr client directly:

import { FlareSolverrClient } from "@/client.js";

const flaresolverr = new FlareSolverrClient({
  baseUrl: "http://localhost:8191",
  defaultTimeout: 60000,
});

// Check health
await flaresolverr.health();

// Create a session for persistence
const sessionId = await flaresolverr.createSession();

// Solve URLs within the same session (cookies persist)
const result1 = await flaresolverr.solveUrl("https://site.com/page1", { sessionId });
const result2 = await flaresolverr.solveUrl("https://site.com/page2", { sessionId });

// Clean up
await flaresolverr.destroySession(sessionId);