Quick Start
Find a simulator, read the current UI, tap a control, type text and capture the result.
1. Find a target
axe list-simulatorsCopy the UDID of a booted simulator and keep it in your shell:
export UDID=<UDID>
axe doctor --udid "$UDID"For a physical iPhone or iPad, run axe list-devices and use the hardware UDID it prints. Every command below works the same way on a device.
2. Read the screen
axe describe-ui --udid "$UDID" # the whole tree as JSON
axe describe-ui --point 200,400 --udid "$UDID" # the element at one pointEach element has fields such as AXUniqueId, AXLabel, AXValue, type and frame:
{
"type": "Button",
"AXLabel": "Continue",
"AXUniqueId": "ContinueButton",
"frame": { "x": 96, "y": 642, "width": 201, "height": 50 }
}Use AXUniqueId or AXLabel values as selectors when you can.
3. Tap
axe tap --id ContinueButton --udid "$UDID"
axe tap --label "Continue" --wait-timeout 5 --udid "$UDID" # poll until it appears
axe tap -x 200 -y 400 --udid "$UDID" # a pointtap prints a ✓ Tap at … completed successfully line when the input was sent. --wait-timeout is the normal way to handle UI that appears after navigation or an animation.
4. Type
axe tap --id EmailField --udid "$UDID"
axe type 'cam@example.com' --udid "$UDID"
axe key 40 --udid "$UDID" # Returntype uses the US keyboard layout. A character it cannot send, such as é, fails with invalidArguments before anything is typed.
5. Capture the result
axe screenshot --udid "$UDID" --output after-login.pngscreenshot prints the path of the PNG it wrote.
6. Run several steps at once
axe batch --udid "$UDID" \
--step "tap --id EmailField" \
--step "type 'cam@example.com'" \
--step "key 40"Using an AI agent?
Agents should use agent mode, which checks each action against the screen and returns the new screen in the same reply:
axe describe-ui --agent --view navigation --udid "$UDID"
axe tap --agent --snapshot-id "$SNAPSHOT_ID" --ref e9 --udid "$UDID"Install the skill with axe init and read Agent Mode.