Browser automations

Browser automations let xopc repeat website actions. Record your own actions, or create and test reusable steps with the assistant. You describe the result in everyday language; xopc handles the browser steps and saves the reusable version.

You do not need to write YAML, selectors, or automation code.

Browser automation or Automation?

These two features work together, but solve different problems:

Feature	What it controls	Example
Browser automation	What the browser does	Open an order page, enter an order number, and return its status
Automation	When the task starts	Run that order lookup every weekday at 09:00

Create a browser automation first when you want to reuse a sequence of actions on a website. Add it to an Automation only when it also needs a schedule or another trigger.

Record your own actions

Use a Chrome extension build with recording support and a compatible Gateway. Update and connect both, then grant the target site access through the extension guide. Update an older extension if its recording panel is missing.

1. Open Browser recording in the extension side panel and select the target website tab.
2. Sign in first, then choose Start recording and perform a small task, such as entering an order number and looking up its status.
3. Use Pause for actions you do not want recorded and Resume when ready.
4. Choose Finish and save, wait for event synchronization, and confirm that the saved automation appears.
5. Inspect its input parameters, enter a test value, choose Run, and check the actual website result.

Recorded fill and select values become reusable inputs and may include the original values as defaults. Do not record passwords, verification codes, payment details, or other sensitive input. Saving confirms stored steps; it does not establish a verified replay.

Recording covers common clicks, fills, selections, checkboxes, key presses, and navigation. Not every website control is supported. If events are missing, a target is unsupported, or recording is interrupted, inspect the error and record a smaller task. If a run's outcome is uncertain, inspect the website before repeating an external write.

Create one with the assistant

1. In the Gateway console, open More apps → Browser automations.
2. Select Create with assistant.
3. Tell the assistant what result you want and which website to use.
4. Complete the task together once. Sign in or confirm sensitive steps yourself when needed.
5. After the task succeeds, the assistant saves the working steps and any values that should be entered on future runs.

For example:

Open my order management page, look up an order number, and return its current status. After it works, save it as a browser automation and make the order number an input.

The first successful run matters: it gives the assistant a real page and a verified path to save. If the website or the desired result is unclear, the assistant may ask a short follow-up question.

Run it again

Open More apps → Browser automations, select an enabled automation, complete the fields under Before running, and select Run now.

The page shows the current status, result, and recent activity. You can stop a run that is still in progress.

You can also ask in chat:

Run my order status lookup for order 12345.

The assistant finds the matching enabled browser automation, supplies the inputs, waits for it to finish, and returns the result.

Edit, pause, or delete

• Select Edit with assistant to explain what should change. The assistant tests the new steps before saving them. Earlier run records retain the exact version used for those runs.
• Select Pause to prevent new runs without deleting the automation. Paused items cannot be selected by scheduled Automations.
• Select Enable when it is ready to use again.
• Select Delete automation when you no longer need it. Deletion is permanent and is unavailable while the automation is running.

Run it on a schedule

1. Make sure the browser automation is enabled.
2. Open Automations and create a new automation.
3. Choose a manual, scheduled, or webhook trigger.
4. Choose Browser automation as the action.
5. Select the saved browser automation and complete its input fields.

The xopc Gateway and its configured browser must be available when a scheduled run starts. A login session may expire between runs, so account-based tasks can occasionally require you to sign in again.

Login and sensitive actions

• Sign in through the browser yourself. Do not save passwords, one-time codes, recovery codes, or payment details as automation inputs.
• Complete CAPTCHA, two-factor authentication, consent, and other human verification yourself. Browser automations do not bypass these controls.
• Check the risk label before running. xopc identifies automations that only read information, change account data, or perform a sensitive action.
• Review website changes carefully. A redesigned page can make previously saved steps fail or behave differently.

Responsible use

Only automate websites, accounts, and data that you are authorized to use. Follow the website's terms, robots and access policies, and reasonable request limits.

xopc restricts each saved browser automation to its declared websites, including redirects. It also blocks unsafe URL patterns. Do not use browser automations to bypass access controls, CAPTCHA, rate limits, paywalls, or anti-bot protections. If a website rejects automation, stop the task and use its official API, export, or manual flow instead.

xopc does not currently ship a catalog of third-party browser automations. This keeps the saved behavior tied to a task you performed and verified in your own environment.

Common problems

Problem	What to do
I cannot find the page	Open More apps → Browser automations, or go to #/browser-automations.
The assistant says browser access is unavailable	Ask the owner of this xopc instance to enable browser access for the selected Agent and install the browser runtime.
The automation asks me to sign in	Open the browser, sign in yourself, then run it again. Never place the password in a reusable input.
A saved automation is missing from Automations	Enable it on the Browser automations page first.
A field is rejected	Complete every required field using the type and format shown on the page.
It used to work but now fails	The website may have changed. Select Edit with assistant and test the task again.
A scheduled run did not start	Check that the Gateway and browser were running, the automation was enabled, and its login session had not expired.

For extension installation, permissions, and connection diagnostics, see the Chrome extension guide. The instance owner can also run xopc browser doctor.

Related guides

• Automations — run work manually, on a schedule, or from a webhook
• Chrome extension — install, connect, grant site access, and troubleshoot the browser bridge
• Tools — configure browser access for an Agent

Execution permissions

Browser control does not ask for confirmation on every click. Recording operates a real website, and replay can change real data. Test a small scope first and define boundaries for sending, deleting, purchasing, or account changes. A risk label does not replace checking the actual actions.