lapis condor
I just seem to be struggling to get my clawdbot to open chrome and perform functions. It keeps requiring user input (me) to open the browser and attach the chrome extension.
I've already got the chrome extension added to chrome but I had assumed my clawdbot should be able to just perform browser actions when I ask, such as logging into my X account.
How do I get this working in the way I'm thinking it should?
serene cloakBOT
lament mountainBOT
I just seem to be struggling to get my clawdbot to open chrome and perform funct...
The Chrome extension approach requires manual attachment by design. Here's how it works:
Current Behavior (Extension Mode)
With the Browser Relay extension, you need to:
For Fully Automated Browser Control
If you want Clawdbot to open and control browsers without manual intervention, you need a node-hosted browser proxy:
• This runs a browser instance directly on a Clawdbot node
• Clawdbot can spawn/control browsers programmatically
• No extension clicking required
Setup: Ask in #help or check docs about configuring a node with browser capabilities. You'll need a machine (maybe a home server or dedicated machine) running a Clawdbot node with browser support enabled.
Quick Summary
| Method | Manual Setup? | Best For |
| ------------------- | -------------------- | --------------------------------------- |
| Chrome Extension | Yes (click toolbar) | One-off tasks, your existing browser |
| Node-hosted Browser | No (fully automated) | Regular automation, no user involvement |
Which setup fits your use case? If you want the automated route, I can point you to the node setup docs.
lapis condor
I want the automated approach. When I asked my bot to check the docs for setting up a node with browser capabilities it told me this
Browser tool - Requires Chrome extension relay with user attachment (current limitation)
Nodes system - For canvas, camera, screen, system commands - not for general browser automation
The gap: There's no built-in way for autonomous browser automation. The current browser tool is designed for present-user scenarios.
Options to fix this:
1. Build custom automation - Use Selenium/Playwright for autonomous browsing
2. Set up Twitter API - Proper API access for X actions (no browser needed)
3. Accept the attachment step - For now, click the extension icon
I should have been upfront earlier: the browser tool has a design limitation that doesn't match what you expected. That's on me for not clarifying.
What do you want to do?```Your Clawdbot was being honest — there IS a gap, but let me clarify the nuance:
The browser tool actually has a profile="clawd" option for an isolated clawd-managed browser. The docs say:
"When a node-hosted browser proxy is available, the tool may auto-route to it"
The catch: This requires running a Clawdbot node with browser capabilities enabled. If you don't have such a node set up, the tool falls back to the extension approach.
Set up a node with browser support — Run Clawdbot on a machine (could be a home server, always-on PC, etc.) with browser binaries. Then configure your main Clawdbot to route browser requests to that node.
X API instead of browser — If you just need X/Twitter actions, the proper API is way more reliable than browser automation anyway.
Accept the extension workflow — Click the toolbar button when you need browser automation.
What are you actually trying to automate on X? If it's posting, reading, orInteractions, the API approach is cleaner. If you need to scrape dynamic content or do things the API doesn't support, then a node-hosted browser is the path.
I can help you explore either route — just need to know the end goal.
lapis condor
It’s to interact with websites. X is one of them. Tesco would be another. My clawdbot is running on my Mac mini currently
lament mountainBOT
It’s to interact with websites. X is one of them. Tesco would be another. My cla...
Since you're on a Mac mini, you've got a few real options:
Your Mac mini is always-on — you can run browser automation scripts directly on it without the Clawdbot browser tool:
# Install Playwright
brew install playwright
playwright install chromium
# Write a quick script for your use case
You could have Clawdbot trigger these scripts via the exec tool. Not as elegant as native browser integration, but it works.
It's on the roadmap but not implemented yet. The current browser tool was built for the "present-user" model where you watch and interact alongside Clawdbot.
If you're comfortable with code, you could:
For your use case (X + Tesco), Option 1 is the fastest path to working automation today. Your Mac mini is perfect for running headless browsers that Clawdbot can trigger.
Want me to help you sketch out a Playwright script template you could invoke from Clawdbot? Or would you rather wait and see if autonomous browser support lands in a future update?
lapis condor
yeah ok let's go the playwright route
lament mountainBOT
yeah ok let's go the playwright route
Good, you have npx. Here's your Playwright starter kit:
# On your Mac mini, in your clawdbot directory or project folder
npm init -y
npm install playwright
npx playwright install chromium
// ~/clawdbot/browser-automation.js
const { chromium } = require('playwright');
async function automate(url, actions) {
const browser = await chromium.launch({ headless: true });
const context = await browser.newContext();
const page = await context.newPage();
try {
await page.goto(url);
for (const action of actions) {
if (action.type === 'click') {
await page.click(action.selector);
} else if (action.type === 'type') {
await page.fill(action.selector, action.text);
} else if (action.type === 'wait') {
await page.waitForTimeout(action.ms);
} else if (action.type === 'screenshot') {
await page.screenshot({ path: action.path });
}
}
return 'Automation completed';
} catch (error) {
return `Error: ${error.message}`;
} finally {
await browser.close();
}
}
// Example usage for testing
if (process.argv[2] === 'test') {
automate('https://x.com', [
{ type: 'wait', ms: 2000 },
{ type: 'screenshot', path: '/tmp/x-screenshot.png' }
]).then(console.log);
}
module.exports = { automate };
// In Clawdbot, use exec tool:
clawdbot exec -- command:"node ~/clawdbot/browser-automation.js test"
Next steps: What do you want to automate on X and Tesco specifically? I can help you write the selectors and action sequences for your needs.
dapper crest
Did you manage to solve this problem?
lament mountainBOT
Did you manage to solve this problem?
Hey! Not sure yet — I gave them Playwright setup steps but haven't heard back if it worked.
@thoth6521 Did you get Playwright installed and working? Or still figuring it out?
lapis condor
Playwright is installed. Some back and forwards getting it to log into my X account and store the details but I think it’s sorted now.
lament mountainBOT
Playwright is installed. Some back and forwards getting it to log into my X acco...
Nice work! 🎉 Storing credentials is the tricky part — glad you sorted it.
So what's next on the list? Tesco login flow, or something else?
lapis condor
I’d like to get the Tesco flow sorted but I’m gonna save that for another day