Hey Prompt Warrior,
Claude Code can now control your browser. Like, actually click buttons, fill forms, navigate between pages, and post things on your behalf.
This means you can automate pretty much anything that happens inside a browser. And since most of your daily work probably lives in a browser, that's a big deal.
In this post, I'll walk you through the 3 different browser automation tools I use, when to pick each one, and I'll share real examples from my own setup.
If you prefer watching over reading, here's the YouTube video.
Here's what we'll cover:
Claude in Chrome for tasks that need your logged-in browser
Playwright CLI for background automations that run on a schedule
Playwright MCP for visual research and multi-step exploration
Real examples: YouTube comment pinner, Skool auto-poster, competitor funnel teardown
Read time: 5 minutes
📣 Quick plug
If you're an agency owner, coach or consultant doing $10K-$100K/mo, I'd love to help you deploy AI agents in your business.
What you get:
Work 5x faster and save 30+ hours/week
Run your business with 2 people doing the work of 10
Install AI systems that actually compound
Spots are limited. If that sounds like you, book a call here.
Now, back to the tutorial.
🔌 Why Browser Automations?
Quick context on when browser control actually makes sense.
If a tool has a CLI, MCP, or API, use that. Those are optimized for agents, they're fast, and they're reliable. Browser automation is the fallback for everything else.
But here's the thing: a lot of tools don't have any of those. Your Skool community, your YouTube comment section, your competitor's landing page, most CRMs, most dashboards. For all of these, the only way in is through the browser.
And now Claude Code can handle that for you.
🖥️ Tool 1: Claude in Chrome (Your Real Browser)
This connects Claude Code directly to the Chrome window you already have open. All your logged-in sessions, cookies, saved passwords, everything is already there.
Best for: tasks that need your existing logins. Posting a comment from your YouTube account, interacting with Notion, updating your outdated CRM.
How to Set It Up:
Open Claude Desktop
Go to Customize > Connectors
Make sure "Claude in Chrome" is enabled (should be on by default) and click to install the extension.
That's it. Now you just create a skill describing the steps.
Real Example: YouTube Comment Pinner
I built a skill that pins a CTA comment on every new YouTube Short I post. Here's what it does:
Opens YouTube in my browser
Checks it's logged into the right profile (switches if not)
Navigates to my channel, clicks the Shorts tab
Opens the latest video
Posts and pins a comment with my CTA
9 steps total. Took about 10 minutes to build. I turned it into a skill and now I just trigger it whenever I post.
Pro Tip: Describe each step clearly when creating the skill. Click through the flow yourself once first so you know exactly what it should do.
⚙️ Tool 2: Playwright CLI (Background Automations)
This runs a separate browser that Claude controls. You log in once, it saves the session, and from then on it can run headless (invisible, in the background).
Best for: repeatable workflows you want to schedule. Post to Skool every time you publish a video. Run a daily dashboard check. Anything that should just happen without you watching.
How to Set It Up:
Tell Claude Code to install
playwright-cliRun the browser in visible mode, log in once
Every future run reuses that saved session (headless or visible)
Real Example: Skool Auto-Poster
Every time I publish a longform YouTube video, this automation:
Grabs the latest video from my Prompt Warrior channel
Opens Skool via Playwright CLI in the background
Writes a post with the video title and embedded link
Selects the right category
Publishes it
I don't see anything on screen. It runs headless. The post just appears in my community.
The best part: you can turn any Playwright CLI workflow into a skill, then schedule it as a routine. So it fires automatically, in the background, on whatever schedule you set.
🔍 Tool 3: Playwright MCP (Visual Research)
This gives Claude full visual context of the page. It takes screenshots, reasons about what it sees, and makes decisions step by step.
Best for: multi-step research where each action depends on what the previous one found. Competitor funnel teardowns, landing page audits, signup flow analysis.
How to Set It Up:
Tell Claude Code to install the Playwright MCP
It spins up its own browser window and works through pages visually
I pointed the Playwright MCP at a competitor's website and asked it to walk through their entire newsletter signup flow.
It navigated the homepage, found the lead magnets, analyzed the signup forms and CTAs, and gave me a structured teardown with specific things I could apply to my own funnel.
This kind of multi-step visual reasoning is where the MCP shines. The CLI would've struggled here because it doesn't "see" the page the same way.
🧭 The Decision Tree
Here's how I pick which tool to use:
Needs your existing browser logins → Claude in Chrome
Should run in the background / on a schedule → Playwright CLI
Visual research across multiple pages → Playwright MCP
Tool has a CLI, MCP, or API → Skip browser, use that instead
🎯 TLDR
✅ Claude Code can now control your browser with 3 different tools
✅ Claude in Chrome = your real browser, your logins, instant access
✅ Playwright CLI = background automations, saved sessions, headless mode
✅ Playwright MCP = visual research, multi-step page exploration
✅ Browser automation is the fallback when no CLI/MCP/API exists (which is a lot of tools)
✅ Build a skill describing the steps, schedule it, stop clicking buttons yourself
Thanks for reading!
If you enjoyed this, consider forwarding this newsletter to a friend or colleague.
What would you like me to write about in the next post? Reply to this email and let me know.
Moritz
P.S. Want help setting up browser automations (and the full ClaudeOS system) for your business? Check out AgentOS. We install these systems so you don't have to figure it all out yourself.

