Skip to content
Anslation

Browser automation glossary, in plain words.

This browser automation glossary covers the terms you meet when you automate a browser, each defined in a sentence or two, with a note on how AnsBot uses it. Every term has its own link.

Every term, in plain words. Grouped by what you do with it.

Building a bot

Browser automation

Software that drives a web browser to repeat a task: open pages, fill in forms, click and read the result.

In AnsBot: a bot is an ordered list of steps.

Step

One action in a bot, such as opening a page, filling a field or reading a table.

In AnsBot: there are 12 step types.

Selector

A pattern that points at elements on a page, such as a CSS selector.

In AnsBot: steps can also find elements by role, label, placeholder or visible text.

Input

A named value a bot fills in at run time, written {{name}} in a step.

In AnsBot: inputs come from the console, a schedule, a webhook body, a batch row or a portal.

Recorder

A browser extension that turns what you do on a page into steps.

In AnsBot: the AnsBot Recorder does not send what you type.

Draft

A bot or version that is not live yet.

In AnsBot: bots that AnsBot drafts for you are saved as drafts and never run on their own.

Version

A numbered snapshot of a bot's steps.

In AnsBot: versions are draft, published or retired.

CSS selector

A pattern that points at page elements by tag, class, id or attribute, like #email.

XPath

A way to select an element by its path through a page, for when a CSS selector cannot reach it.

DOM (Document Object Model)

The live tree of elements a browser builds from a page, which a step reads and acts on.

Locator

How a tool finds an element at run time: a selector plus rules like a role, a label or visible text.

Running a bot

Headless browser

A browser that loads pages without showing a window.

In AnsBot: runs use headless Chromium or Firefox on Anslation's servers.

Run

One execution of a bot, start to finish.

In AnsBot: a run is queued, running, succeeded, failed, needs human or cancelled.

Test and production mode

Two ways to run a bot: to try it, or to do the real work.

In AnsBot: a permission policy applies to production runs unless you include test runs.

Environment

A named set of run settings, such as test, staging or production.

In AnsBot: a run you start with an environment uses its version, inputs, sign-in and browser; schedules, webhooks and batches carry their own settings.

Data batch

A list of records, each run through the same bot.

In AnsBot: up to 1,000 rows per batch, one run per row.

Duplicate guard

A rule that stops the same record being processed twice within a time window.

Dead-letter queue

Where work goes after it has failed every retry, to wait for a person.

In AnsBot: an operator can requeue it.

Checkpoint

A saved point in a run that lets it resume from a step instead of starting over.

Browser session

The cookies and stored data that keep you signed in to a site while a bot works.

Retry

Running a failed step or run again, sometimes after a wait, in case the cause was temporary.

Flaky

An automation that passes sometimes and fails other times without the task itself changing.

Oversight and evidence

Human in the loop

A design where automation stops for a person's decision before it carries on.

In AnsBot: steps you mark wait for approval.

Approval request

A request for a person to approve or reject a step a run is waiting on.

In AnsBot: it carries the reason and a screenshot of the page.

Permission policy

Rules for which actions and domains a bot may use.

In AnsBot: when switched on, it asks for approval before submit-like clicks, uploads, stored sign-ins and moves to another domain.

Step log

The record of what each step did.

In AnsBot: status, the element used, a message and the output.

Failure trace

A recording of a failed browser session that shows each action and the page at that moment.

In AnsBot: kept privately for a limited time, for failed or waiting attempts.

Self-healing

Automation that finds a new way to an element when a page changes.

In AnsBot: repairs are tested in a shadow run and approved by a person.

Shadow run

A test run of a changed copy of a bot, with the same inputs, before the change goes live.

Vault

Encrypted storage for the sign-in details a bot uses.

In AnsBot: values are shown masked and redacted in logs.

Two-factor authentication

A second sign-in step, such as a code from an app or a device prompt, on top of a password.

Connecting and delivering

Webhook

An address another system calls to start something.

In AnsBot: a POST with the endpoint's token queues a run.

Callback

A request a service sends back to you when work finishes.

In AnsBot: signed with HMAC-SHA256 in the X-Ansbot-Signature header when the endpoint has a signing key.

Client portal

A page where a client sees and uses work done for them.

In AnsBot: a signed, expiring link to start runs and export results.

Smoke check

A quick check that the most important pages of a site still work after a change.

API

A defined way for programs to exchange data directly, without opening a browser or a page.

Read next. Put the terms in context.

Glossary: questions.

Still have a question? Ask us

What does "headless" mean in browser automation?

That the browser runs without a window. It loads and runs pages the same way, which is how automation runs on servers.

What is the difference between a selector and a locator?

A selector is the pattern, such as a CSS selector; a locator is how a tool uses it, often with extra rules such as a role, a label or visible text.

Where do these terms come from?

From how browser automation is built and run in general, with a note on how AnsBot uses each one.

Seen enough terms? See them at work in AnsBot.

Tell us the web task and the site it runs on. We read every request and reply by email.

Office hours: Monday to Saturday, 10:00–18:00 IST+91 74158 85838