1. Docs
  2. Reference
  3. Step reference

Step reference

A test case is a list of steps. This reference lists every step type SmartestQA offers, which project types have it, where you add it, and what happens to the run when it passes or fails.

10 min read Updated CustomersAdmins

In short

On this page you can

  • Find the right step type for a task, from a click to a database query.
  • See which step types exist in WEB, MOBILE and DESKTOP projects.
  • Learn which failures stop a test case and which only mark the step.
  • Look up the variables each step produces and how later steps read them.
  • Open the family page with fields, examples and errors for each step.

How to use it

  1. Open a test case in Automate and look at the step toolbar above the step list.
  2. Pick a button such as Action, Assert or API, or open More for the rest.
  3. Fill the step’s dialog, then select the step to set its values in the right panel.
  4. Pass values between steps with {{tc:name}} variables.
  5. Click Save, then Dry Run to see each step’s result in the live log.

At a glance

Where
Automate › a test case › the step toolbar above the step list. Less common types are under More.
Who
Every user of the project can add, change and remove steps. There is no role check on steps.
Two kinds
Action steps point at an action from Discover (a control on a page). Step-type steps (Assert, Python, API, Mail…) are created by their own dialog.
Values
Each step has parameters in the right panel. Any value can contain a variable such as {{tc:orderId}}.
Results
Each step ends as SUCCESS, FAILED or ERROR (Network steps can also end as WARNING).
Runs in
Dry Run from the editor, pipelines in Schedule, on the server or on your agent. The same steps run in all of them.
Project types
WEB, MOBILE (Android and iOS) and DESKTOP (Windows and macOS) each have their own action types. Assertion, Python, API, Database, Mail, Kafka, Global parameter, Wait and AI exist in all three. JavaScript and Network are web only. JMeter and k6 are web and desktop.

Start from your question#

Where steps fit#

Steps are built in Automate, the third stage of the order of work. Action steps reuse the actions you captured in Discover; pipelines in Schedule run the finished test cases.

  1. 1DiscoverActions and pages
  2. 2AnalyzeRequirements
  3. 3AutomateSteps in test cases
  4. 4SchedulePipelines run them
  5. 5ReportStep results
  6. 6MaintenanceHeal and review

For the editor itself (creating test cases, the step list, saving, Dry Run), see Automate. To capture actions first, see Discover › Actions.

All step families#

The toolbar shows the everyday types. Everything else is in More, grouped under Basic, Integrations and Load test. When the toolbar is too narrow, buttons that do not fit move into More as well.

FamilyStep typesProject typesAdded from
Browser actionsClick, Double Click, Right Click, Hover, Type, Clear & Type, Select, Navigate, Get, Visibility Check, Wait (seconds)WEBAction, Record
Mobile actionsTap, Swipe, Scroll, Type Text, Launch App, Press Back, Get Text, Wait For Element, Assert Text and 25 moreMOBILEAction, Global, the Appium Inspector
Desktop actionsClick, Type Text, Type Key by Key, Press Key, Select, Switch Window, Assert Enabled and 24 moreDESKTOPAction, Global, the Desktop Recorder
Checks and assertionsAssertion (Assert), Visibility Check, AI Check, Network Check, mobile and desktop Assert actionsAllAssert
Waits, conditions and templatesWait (seconds), Wait For… actions, IF / ELIF / ELSE / END, TemplateAllAction, Python, Template
Data and variablesGlobal parameter, Get, step outputs, data sourcesAllMore › Global parameter
CodePython, JavaScriptPython: all. JS: WEBPython, JS
AI stepsAI Ask, AI Check, AI Do, AI ChatAllAI step
APIAPI RequestAllAPI
NetworkCheck, Get value, List (browser calls)WEBNetwork
DatabaseDB (SQL query)AllMore › Database
MailGet Mails, Extract OTP, Extract Link, Send Mail, Wait for MailAllMore › Mail
KafkaKafka Produce, Kafka ConsumeAllMore › Kafka
Load testsJMeter, k6WEB, DESKTOPMore › JMeter / k6
Automate step toolbar of a web test case with the More menu open, showing the Basic, Integrations and Load test sections
The step toolbar and the More menu.

The building blocks#

Step
One line in a test case. It runs in order, from step 1 down. A step has a name, a type, and parameters.
Action step
A step that points at a shared action (for example Click Sign In on page login). Fixing the action’s locator fixes every test that uses it. The step list shows how many test cases use the action.
Step-type step
A step created by its own dialog: Assertion, Python, JavaScript, AI, API, Network, Database, Mail, Kafka, Global parameter, JMeter, k6. These are not in the Action List and belong to one test case.
Parameter
A key and a value on the step, shown in the right panel under Parameters List. For a Type step the value is the text to type; for a Get step the key is the name of the variable to fill.
Variable
A value saved by one step and read by a later one, written {{tc:name}}. See Data and variables.
Output
The variable a step writes: the Get key, an API JSON path, SQL output values, a mail OTP, a code step’s returned value.
Template
A test case marked as a template, added inside another case as one step. Its steps run in place. See Templates.

Step results#

Every step reports one status in the live log of a Dry Run and in the report of a pipeline run.

StatusMeaningTypical causes
SUCCESSThe step did what it should.Element clicked, assertion met, query returned rows.
FAILEDA check ran and did not pass.Assertion not met, Visibility Check false, Network check failed, AI Check answered false, load test with failed requests.
ERRORThe step could not run.Element not found, connection refused, wrong SQL, code exception, missing configuration. Reports store it as failed.
WARNINGNetwork step only: failed calls or JS errors were seen and On failed calls / JS errors is Mark as warning.A 500 response in scope.
SKIPPEDLive log only: the step did not run because an earlier step stopped the case.Any stopping failure above it.

Which failures stop the test case#

Most failures stop the test case: the remaining steps do not run. A few step types only mark themselves and let the case go on. Plan your checks with this table.

StepWhen it goes wrongThe case…
Click, Type, Select, Hover and other element actionsElement not found after retries and auto-healstops
AssertionCondition not met, or configuration errorstops
Visibility CheckElement or text not visiblecontinues (step FAILED)
API RequestAny HTTP status, including 4xx and 5xxcontinues (step SUCCESS)
API RequestTimeout, SSL or connection errorcontinues (step ERROR)
Database, Mail, KafkaConnection, query or timeout errorstops
NetworkCheck fails, or no matching call with Fail the stepstops
Python, JavaScriptException, or STATUS output is falsestops
AI stepAI error or a failed AI Checkstops for steps added from the dialog (they are created with stop-on-error); continues otherwise
JMeter, k6Failed requests, timeout, tool missingthe step ends as ERROR in the report

Check API results yourself

An API step passes on any HTTP answer. Save @status into a variable and add an Assertion after it. See Check the status code.

Where steps run#

A test case runs on the server (headless, no agent needed) or on your agent (the SmartTest app on your computer). Steps that connect to other systems connect from that machine.

What the step reachesConnects fromWatch out
Your app’s UI (web, mobile, desktop)The browser, device or desktop session of the runMobile runs need a device or emulator; desktop runs need an agent on the Windows or macOS computer with the app.
API, Database, Mail, KafkaThe machine that runs the testA database in your office network is not reachable from the server. Run on your agent.
AI stepsThe project’s AI providerAI must be enabled for the project.
JMeter, k6The machine that runs the testJMeter must be installed there. k6 is downloaded automatically unless that is turned off.

Differences between a Dry Run and a pipeline run that matter for steps:

Best practices#

Do

  • End every flow with a check: an Assertion, a Visibility Check or an AI Check.
  • Use Clear & Type for fields that already contain text.
  • Name variables after what they hold: orderId, not value1.
  • Put a login or setup sequence in a template and reuse it.
  • Assert on @status after every API step that must succeed.
  • Run database, mail and Kafka steps on an agent inside your network.

Don’t

  • Add fixed waits to fix timing. Use a Wait For… step or a check that waits.
  • Rely on Visibility Check to stop a broken flow; it lets the case continue.
  • Leave AI steps in place once a regular action can do the job.
  • Write UPDATE or DELETE queries against shared data without a WHERE clause.
  • Mix JMeter or k6 with browser steps; load tests need their own test case.

In this section#

Step referenceBrowser actions

The web action steps: Click, Double Click, Right Click, Hover, Type, Clear & Type, Select, Navigate, Get, Visibility Check and Wait.

Step referenceChecks and assertions

The Assertion step and its types, how values are compared, and which check to use when: Assertion, Visibility Check, AI Check, Network Check and the mobile and desktop Assert actions.

Step referenceWaits, conditions and templates

Fixed and conditional waits, how long a step waits for its element, IF / ELIF / ELSE / END blocks, and template steps that reuse a whole sequence.

Step referenceData and variables

Pass values between steps: the Global parameter step, the Get step and its transforms, which steps produce variables, and the {{tc:}}, {{ds:}} and {{excel:}} reference syntax.

Step referenceCode steps: Python and JavaScript

Write Python or JavaScript in a step, return a value or a pass/fail result, insert ready commands, and build IF / ELSE blocks.

Step referenceAI steps

Ask about the screen, check something visible, act in plain words, or chat with the AI as a step: the four AI modes, their fields and outputs.

Step referenceAPI Request step

Call an HTTP API inside a test case, try it with Send, save response values with JSON paths, and check the status code.

Step referenceNetwork step

Check, read or list the calls the browser makes during a web test: status codes, response fields, durations, JS errors.

Step referenceDatabase step

Run an SQL query against your database inside a test case and save column values as variables.

Step referenceMail steps

Read a one-time code or a link from an email, wait for a message, count messages, or send an email during a test.

Step referenceKafka steps

Publish a message to a Kafka topic and wait for a matching message, to test event-driven flows inside a test case.

Step referenceLoad test steps: JMeter and k6

Add a JMeter or k6 load test as a step: upload a .jmx or .js file, or generate one with AI from the API steps of a test case.

Step referenceMobile actions

Every Android and iOS action step: touch and gestures, input, app lifecycle, navigation, reading values, waits and assertions, plus the ready-made Global actions.

Step referenceDesktop actions

Every Windows and macOS action step: mouse, input and keys, scrolling, waits, assertions, reading values, app and window control, plus Global desktop actions.

Step referenceStep troubleshooting

Every step-level message and surprise, from element not found to mail timeouts, with why it happens, what to check and how to fix it.

Didn’t find your answer?Write to support with your project name and what you see on screen.

info@smartestqa.ai
© 2026 Selenity Technology · SmartestQAsmartestqa.aiThis page as MarkdownChecked against smarttest-fe predev 2a8c79cc, smarttest-be 633c2a03, smarttest-testsuite 017fcec