Commands

Fetch

Fetch one or more URLs and dump the rendered result to stdout. Each URL is loaded in its own page, with JavaScript executed, so you get the final DOM rather than the raw server response.


lightpanda fetch --obey-robots --log-level info --dump html https://demo-browser.lightpanda.io/campfire-commerce/


$time=1785335542796 $scope=frame $level=info $msg="navigate" url=https://demo-browser.lightpanda.io/campfire-commerce/ method=GET reason=address_bar body=false req_id=1 type=root
$time=1785335543068 $scope=browser $level=info $msg="executing script" src=https://demo-browser.lightpanda.io/campfire-commerce/script.js kind=javascript cacheable=true
$time=1785335543139 $scope=http $level=info $msg="request complete" source=xhr url=https://demo-browser.lightpanda.io/campfire-commerce/json/product.json status=200 len=4770
$time=1785335543140 $scope=http $level=info $msg="request complete" source=fetch url=https://demo-browser.lightpanda.io/campfire-commerce/json/reviews.json status=200 len=1615
<!DOCTYPE html>
...

Choose the output format with --dump and control how long the page runs with the --wait-* options. Find all options in the fetch command reference.

Serve

Start Lightpanda as a Chrome DevTools Protocol (CDP) server to control it with clients like Puppeteer , Playwright or Chromedp .


lightpanda serve --obey-robots


$time=1785335600862 $scope=app $level=note $msg="server running" address=127.0.0.1:9222

Find all options in the serve command reference.

Connect a CDP client once the server is running:

MCP

Start Lightpanda as a Model Context Protocol (MCP) server over stdio to control it with LLMs.


lightpanda mcp

Find all options in the mcp command reference and the available tools in the MCP tools reference.

Connect a MCP client once the server is running, using your preferred LLM. Find details about the MCP usage.

Agent

Start Lightpanda as an interactive AI agent that can browse the web.


lightpanda agent

The agent auto-detects an API key from your environment and opens a REPL where you type instructions in natural language. You can also run a single task and exit with --task "...", or replay a saved script without any LLM call with lightpanda run script.js.

Find all options in the agent command reference.

Run

Run a saved PandaScript (a .js automation script), then exit. No LLM calls, no API key needed.


lightpanda run script.js

Produce a script to replay with lightpanda agent --task "..." --save script.js, or with /save from the agent REPL.

Find all options in the run command reference.

See Agent and PandaScript usage for more details.