Actions

Tap an element by describing it in plain English, and the six ways to describe it so only one element matches.

Tap presses one element on screen. The description identifies the element; Vision AI locates it.

Platforms	Android · iOS
Targets by	Visible text, icon, position, neighboring element, section, color
Best target	The element's exact visible label
Also accepted	Tap, Tap on, Tap the — all the same command
Watch out	Paraphrasing the label is the most common cause of a wrong tap

Prerequisites

• A connected device or emulator

• An open test file

• The target screen reachable by the steps above the tap

Copy this

OPEN_APP com.shopease.android
Wait Until 5 Seconds

Tap on the search icon
Type running shoes in the search field
Tap on the first search result
Wait Until 3 Seconds

Validate that the product detail page is visible
Tap on Add to Cart
Wait Until 2 Seconds
Validate that Item added to cart is visible

Write a tap step

1. Read the element's exact visible label off the screen.

2. Write Tap on <label> using that wording, character for character.

3. Add one qualifier — position, neighbor, section, or color — when the label appears more than once on the screen.

4. Add Wait Until <n> Seconds after the tap when it triggers navigation or a network call.

5. Add a Validate on the next line when the tap navigates, submits or pays.

6. Run the test and confirm the report shows the tap resolving to the intended element.

Targeting strategies

Drizz selects one element out of everything on screen. The description must be satisfied by exactly one element. The six strategies, in order of reliability:

Strategy	Use when	Example
Exact visible label	The element has text	Tap on Login CTA
Icon by common name	The element has no text	Tap the cart icon
Position in a list	The same label repeats down the screen	Tap on the first Add beside product name
Relative to a neighbor	Each row has its own copy of the label	Tap Remove beside Product Details
Within a section	The same label appears in two sections	Tap Add under Snacks header
Visual attribute or ranking	State or a comparison is what distinguishes it	Tap the green Active button · Tap the highest-rated product

Examples

Exact text — the default:

Tap on Login CTA
Tap on Continue Button
Tap on Continue to Payment

Icons, by the name people use for them:

Tap the cart icon
Tap the profile icon
Tap the hamburger icon
Tap the settings icon
Tap the close icon
Tap the back arrow
Tap the search icon

Position, when the label repeats:

Tap on the first Add beside product name
Tap on the last Remove CTA
Tap the cart icon at top right
Tap Buy on the first card

Section and neighbor context:

Tap Add under Snacks header
Tap Remove beside Product Details
Tap Select next to Basic Plan option
Tap on % option under Discount

Color and state, when nothing else separates them:

Tap the green Active button
Tap the red Retry indicator
Tap the highlighted Selected tab
Tap the orange Offer badge

Ranking and comparison — Drizz selects by value, not only by text:

Tap the highest-rated product
Tap the lowest-priced item
Tap the plan with the maximum discount
Tap the restaurant with the most reviews
Tap the item with rating above 4.5

Use the app's exact wording

If the button reads Charge Now, write Charge Now — not "Proceed to Pay", not "the payment button". Paraphrasing is the most common cause of a tap resolving to the wrong element, and it fails intermittently rather than immediately.

Validate after a tap

A tap that navigates, submits or pays is followed by a Validate. Without one, a mis-resolved tap surfaces several steps later, on an unrelated screen.

Tap on Place Order
Wait Until 3 Seconds
Validate that Order Placed is visible

When a tap won't resolve

Add context first — a neighbor, a section header, a position. MAP_ACTION targets elements that have no label and no clean bounding box.

A tap that keeps failing immediately after a Type might be a keyboard problem — see Type.

Common mistakes

What you wrote	What happens
Tap on the payment button when the button reads Charge Now	Vision AI guesses. It resolves to a different control, or fails on some runs and not others
Tap on Add on a screen with five Add buttons	One of them is tapped, and which one can change between runs
Tap on the third product when list order is server-driven	Passes on your device, fails on the next run with different data. Target by label instead
Tap on the blue button in the middle of the screen near the bottom	Long positional prose is weaker than the label. Use the label, add one qualifier
MAP_ACTION Tap on Continue when Continue has a visible label	A deterministic tap traded for a grid guess
A tap immediately after Type, with no wait	The keyboard or a transition can still cover the target and the step reports as undoable

Next

• Type — entering text, and keyboard handling

• Validate — asserting the result of a tap

• MAP_ACTION — unlabeled elements and gestures

---

Last updated: 6 August 2026