Web Browsing Skill

Browse the web using Clawdbot’s built-in browser tool (not curl!).

Setup

Browser must be enabled in Clawdbot:

# Browser is already enabled in your config
clawdbot browser status  # Check status
clawdbot browser start   # Start if needed

Commands

Open a URL

uv run {baseDir}/scripts/web.py open "https://example.com"

Get Page Content

uv run {baseDir}/scripts/web.py get "https://example.com"

Get Specific Element (CSS Selector)

uv run {baseDir}/scripts/web.py get "https://example.com" --selector ".main-content"

Search the Web (DuckDuckGo)

uv run {baseDir}/scripts/web.py search "Python AI agents"
# Output:
# 1. https://result1.com
# 2. https://result2.com

Get Plain Text

uv run {baseDir}/scripts/web.py text "https://example.com"

Take Screenshot

uv run {baseDir}/scripts/web.py screenshot "https://example.com" --output /tmp/page.png

Examples

Browse Hacker News

# Open HN
uv run {baseDir}/scripts/web.py open "https://news.ycombinator.com"

# Get top stories
uv run {baseDir}/scripts/web.py get "https://news.ycombinator.com" --selector ".titleline"

Research a Topic

# Search for information
uv run {baseDir}/scripts/web.py search "multi-agent AI systems research"

# Open first result
uv run {baseDir}/scripts/web.py open "https://arxiv.org/abs/..."

Get Documentation

# Get page text
uv run {baseDir}/scripts/web.py text "https://docs.example.com/api"

Why This Is Better Than Curl

Browser Tool Directly

You can also use the browser tool directly:

# Open URL
clawdbot browser open "https://example.com"

# Get snapshot
clawdbot browser snapshot

# Take screenshot
clawdbot browser screenshot --out /tmp/page.png

# Navigate
clawdbot browser navigate "https://new-url.com"

# Click element
clawdbot browser act --kind click --selector ".button"

Troubleshooting

“Browser not running”

clawdbot browser start

“Browser not available”

Check if browser is enabled in your config.

Slow pages

Increase wait time in scripts or use text command for faster results.