kernel-browser-pools（内核浏览器池）

Browser Pools

Pre-warmed browser instances for fast acquisition. Ideal for high-throughput automation.

Note: Unless otherwise noted, id arguments refer to the browser session ID, not invocation IDs returned by Kernel commands.

When to Use

Browser pools are ideal when you need to:

• Minimize latency for browser acquisition with complex configurations (profiles, proxies, extensions)
• Run many short-lived browser tasks in parallel with consistent startup times
• Maintain consistent browser configuration across multiple sessions
• High-throughput automation where browser creation overhead impacts performance
• Parallel scraping of many pages with identical browser settings

For one-off or long-running browser sessions, creating browsers directly with kernel browsers create may be simpler.

Prerequisites

Load the kernel-cli skill for Kernel CLI installation and authentication.

Create Pool

# Basic pool
kernel browser-pools create --name my-pool --size 5 -o json

# With browser options
kernel browser-pools create --name my-pool --size 10 --stealth --headless -o json

Manage Pools

List Pools

kernel browser-pools list -o json

Get Pool Details

kernel browser-pools get my-pool -o json

Update Pool Size

kernel browser-pools update my-pool --size 10 -o json

Delete Pool

kernel browser-pools delete my-pool --force

Acquire and Release

Acquire a Browser from Pool

kernel browser-pools acquire my-pool -o json

This returns a browser session from the pool instantly.

Release Browser Back to Pool

kernel browser-pools release my-pool --session-id <id>

Use Case: High-Volume Scraping

# Create pool with 20 browsers
kernel browser-pools create --name scraper-pool --size 20 --stealth -o json

# Acquire browsers as needed
for i in {1..100}; do
  SESSION=$(kernel browser-pools acquire scraper-pool -o json | jq -r '.session_id')

  # Use the browser
  kernel browsers playwright execute $SESSION "
    await page.goto('https://example.com/page-${i}');
    return await page.content();
  "

  # Release back to pool
  kernel browser-pools release scraper-pool --session-id $SESSION
done

# Cleanup
kernel browser-pools delete scraper-pool --force