Agent Skills: Scrape data on tab by index

Daemon auto-starts on first use, auto-kills after 3 minutes of inactivity.

Usage

bun scripts/play.ts tabs                         # list open tabs
bun scripts/play.ts [tab] "code"                 # run browser JS on tab (default: first)
bun scripts/play.ts [tab] https://...            # navigate tab
bun scripts/play.ts [tab] Domain.method [json]   # send raw CDP command
bun scripts/play.ts close [tab]                  # close tab
bun scripts/play.ts new [url|code|cdp]           # open new tab, optionally run/navigate

Code runs as browser JavaScript in the page. All browser globals available: document, window, location, fetch, etc. Use return to output a value. Objects are JSON-formatted. [tab] = 1-based index.

Examples

# Scrape data on tab by index
bun scripts/play.ts 2 "return [...document.querySelectorAll('h2')].map(el => el.textContent.trim())"

Raw CDP Commands

# Enable debugger
bun scripts/play.ts "Debugger.enable"

# Set breakpoint
bun scripts/play.ts 'Debugger.setBreakpointByUrl {"lineNumber":42,"urlRegex":".*app\\.js$"}'

# Pause on exceptions
bun scripts/play.ts 'Debugger.setPauseOnExceptions {"state":"uncaught"}'

# Get cookies
bun scripts/play.ts "Network.getCookies"

Network

# Monitor via JS (intercept fetch in page)
bun scripts/play.ts "
const orig = fetch
window.fetch = async (...a) => { console.log('REQ:', a[0]); return orig(...a) }
"