Automate website actions

What I mean by website actions

I don't just mean a single click or a file upload. A bot or AI can do something that simple on its own. By website actions I mean stringing a set of steps together to finish a real task. The kind of task that loops through every row in a sheet, or reads a value on the page and decides what to do next. That branching and that repetition is where the time goes.

None of it has to stay manual. A complex set of actions breaks down into a sequence of small steps, and once it is in steps you can automate it three ways, with no code, with code, or with a Claude skill.

Who this is for

This is for people like me who spend the day in the browser, filling in forms, uploading files, clicking the same buttons in the same order. You understand software, but you don't have to be a coder. On our team it is a mix of coders and non-coders, and we can all automate the same complex tasks.

If your work is a string of website actions you repeat, axiom chains them into one run you trigger again and again, with looping and logic when a task needs to handle every row and every case, not just the first one. If you use something in Chrome and want to automate it, with code or without, this is for you.

How I'd approach it

I'm a no-coder, so my approach leans no-code. To any coders reading, wait, am I wrong? I don't think so. The no-code tool already handles the things that make browser automation hard. It waits for pages to load. It gives you a selector tool to pick the exact element you want. It has webhooks and security built in. That and a lot more. And if you do want code, you can use it inside the no-code tool anyway. So yes, my approach is no-code.

One more thing. Map out your use case first, the task you are trying to finish, before you break it into actions. Get that clear and the rest follows. That is why I would reach for Build with description every time.

Automate website actions from a description

Describe the task in plain words in the Chrome extension and it builds the steps for you. Give it a few short lines, check what it made, and run it. Explore no-code.

Automate website actions in code

Build with code. If you are more interested in the infrastructure and running your own code, this is the right choice for you. Explore code

import { chromium } from "playwright";

const browser = await chromium.connectOverCDP(
  `wss://cdp-lb.axiom.ai/?token=${process.env.AXIOM_API_KEY}`
);

try {
  const context = browser.contexts()[0];
  const page = context.pages()[0] ?? await context.newPage();

  // 1. Log into the site
  await page.goto("https://example.com/login");
  await page.fill("#email", process.env.SITE_EMAIL);
  await page.fill("#password", process.env.SITE_PASSWORD);
  await page.click("button[type=submit]");

  // 2-4. Search each value (these would come from your Google Sheet)
  const queries = ["widget", "gadget", "gizmo"];
  const results = [];

  for (const query of queries) {
    await page.click("#search");
    await page.fill("#search", query);
    await page.keyboard.press("Enter");

    // 5. Scrape every result page, clicking the pager until it runs out
    while (true) {
      await page.waitForSelector(".result");
      const rows = await page.$$eval(".result", els =>
        els.map(el => el.textContent.trim())
      );
      results.push(...rows);

      const next = page.locator(".pager-next");
      if ((await next.count()) === 0 || (await next.isDisabled())) break;
      await next.click();
    }
  }

  console.log(results);
} finally {
  await browser.close();
}

+

Build with a Claude skill

Build no-code or code bots with a skill.

Add the Claude skill and describe what you want. It builds the bot for you, no-code or code, and handles the tricky parts like proxies and schedules.

Install the skillLearn more

What can you automate?

Most things. A few worth knowing up front.

What I'd watch out for

As we covered, the range of actions we perform in a browser is broad, so the use cases for automating web work are broad too. I won't pretend it is all easy. Complexity and a learning curve are both real in browser automation, for any number of reasons. That is why we are building axiom.

Selector issues

"Element not found" means the element is either missing from the page or the selector is wrong. If you can see the element on the page while the run happens, the selector is the problem, and that you can fix.

Cookies

If you log into a site, you either have the run sign in each time, or you store the cookies so the session carries over. Storing the cookies is the method we recommend.

Iframes

To act on anything inside an iframe, turn iframe support on in the selector tool first. Selector tool docs