Swipe, slider & picker precision
Set exact continuous values with swipes, sliders, seek bars and picker wheels.
20 project-authored proposals · taxonomy v1.1 · 0 new benchmark runs. Verifiers and partial-credit rules below are designs, not executed results.
- Typical applications
- Settings · VLC · Calendar pickers
- Class GUI-need label
G3· GUI-only
Task proposals · 20
Expand to inspect the complete original wording and proposed checks. Original proposal text remains English in both interfaces.
R01.11.02.01-T01 Android: set screen brightness slider to 40%
Original instruction
With adaptive brightness off, use the Quick Settings brightness slider to set screen brightness to 40% of its range (±5%). Do not open the Settings app.
- Platforms
and· Android- Applications
- System UI
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
settings get system screen_brightness_float in [0.35, 0.45] after mapping the slider's gamma curve; screen_brightness_mode=0; Settings activity never resumed. adb and root shell are withheld from the agent (screen input only), so settings put is unavailable; the level must be set with the Quick Settings slider.
Proposed partial credit
70% value within ±0.05 of 0.40; 30% adaptive stays off and QS route used.
R01.11.02.01-T02 Android: set media volume to exactly step 7 of 15
Original instruction
Press a volume key to open the volume panel, expand it, and drag the Media slider to exactly step 7 of 15. Ring and alarm volumes must stay unchanged.
- Platforms
and· Android- Applications
- System UI
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
dumpsys audio: STREAM_MUSIC index 7 (max 15); STREAM_RING and STREAM_ALARM equal snapshot values. adb and root shell are withheld from the agent (screen input only), so 'media volume' and key injection are unavailable; the index must be set with the volume panel slider.
Proposed partial credit
70% media index exactly 7 (40 pts if ±1); 30% other streams unchanged.
R01.11.02.01-T03 Android: swipe an onboarding carousel to card 5 and start
Original instruction
Open TourDemo (test app). Its onboarding carousel has 7 cards and no dots or skip button; swipe to the fifth card, which says 'Pick your region', choose 'EU', then swipe to the last card and tap Start.
- Platforms
and· Android- Applications
- TourDemo
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesvis· Visual judgment- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 2 / 5
Proposed verifier
TourDemo state log: region=EU set on card index 4, onboarding_complete=true; overshooting and returning is allowed but each card's dwell is logged for analysis. adb and root shell are withheld from the agent (screen input only), so am start deep links and run-as state edits are unavailable; cards must be swiped.
Proposed partial credit
40% region EU set on correct card; 60% onboarding completed.
R01.11.02.01-T04 Android: date-of-birth picker wheels to 14 March 1987
Original instruction
In FormDemo's registration screen, set the date-of-birth field to 14 March 1987 using the spinner-style picker (day, month and year wheels; keyboard entry disabled) and save the form.
- Platforms
and· Android- Applications
- FormDemo
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
FormDemo DB row dob=1987-03-14; the picker's text-input mode is disabled in the build, so values must come from wheel scrolling.
Proposed partial credit
40% year correct; 30% month correct; 30% day correct and saved.
R01.11.02.01-T05 iOS Clock: set a 6:47 AM alarm with picker wheels
Original instruction
On the iOS 18 Simulator, add an alarm in Clock for 6:47 AM using the hour/minute/AM-PM wheels, label it 'Shuttle', repeat Monday–Friday, and save.
- Platforms
ios· iOS/iPadOS- Applications
- Clock · iOS picker wheel
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & droptype· Text & forms- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Grader XCUITest snapshot of Clock › Alarms lists 6:47 AM 'Shuttle, weekdays' enabled; simulator alarm store (MobileTimer plist) hour=6 minute=47 repeat mask Mon–Fri. Host shell and xcrun simctl are withheld from the agent, so the MobileTimer plist cannot be edited; the alarm must be set with the picker wheels.
Proposed partial credit
50% time exact; 25% label; 25% repeat days.
R01.11.02.01-T06 Android: set an event time by dragging the Material clock dial
Original instruction
In Simple Calendar, create event 'Vendor call' on 13 Nov 2026 starting 14:35 and ending 15:10, setting both times with the circular clock-dial picker (keyboard toggle hidden in this build).
- Platforms
and· Android- Applications
- Simple Calendar · Material time picker
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
touch· Touch gesturesdrag· Drag & dropgeom· Precise geometry- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Calendar DB event with title, date and dtstart 14:35, dtend 15:10 exact (local tz). adb and root shell are withheld from the agent (screen input only), so content insert into the Calendar provider is unavailable; the time must be set on the clock dial.
Proposed partial credit
20% event created on date; 40% start time exact; 40% end time exact.
R01.11.02.01-T07 Android equalizer: five bands to target dB without presets
Original instruction
In Wavelet (headphone EQ), set the 5-band graphic EQ to 60 Hz +4.0, 230 Hz −2.0, 910 Hz 0.0, 3.6 kHz +3.5, 14 kHz −1.5 dB (±0.5 dB each) by dragging band sliders. Do not load or type presets.
- Platforms
and· Android- Applications
- Wavelet
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropgeom· Precise geometry- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Wavelet shared_prefs band gains read by grader; each within ±0.5 dB; preset field unchanged ('Custom'). adb and root shell are withheld from the agent (screen input only), so run-as prefs edits are unavailable; gains must be set with the band sliders.
Proposed partial credit
40% bands 1-2 within ±0.5 dB (20 each); 60% bands 3-5 within ±0.5 dB (20 each).
R01.11.02.01-T08 Android photo editor: exact brightness and contrast slider values
Original instruction
Open Pictures/dock.jpg in the Gallery editor, set Brightness to +25 and Contrast to −10 exactly (slider readouts), leave other adjustments at 0, and save as a copy.
- Platforms
and· Android- Applications
- Gallery editor
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
New copy exists; grader renders the reference with the editor's documented filter at +25/−10 and requires mean abs pixel diff < 1.0; edit sidecar/DB records brightness=25 contrast=−10. adb and root shell are withheld from the agent (screen input only), so file pushes and run-as DB edits are unavailable; the copy must be saved from the editor.
Proposed partial credit
40% brightness exact; 40% contrast exact; 20% saved as copy, original untouched.
R01.11.02.01-T09 Android VLC: scrub to 01:15 ±1 s on a 40-minute file
Original instruction
In VLC, open Movies/lecture_40min.mp4 and seek to 01:15 (±1 s) using the seek bar only (no double-tap skip, no jump-to-time dialog), then pause.
- Platforms
and· Android- Applications
- VLC
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropav· Audio/video over time- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
dumpsys media_session: VLC state PAUSED, position 75±1 s. VLC input log shows no jump-to-time dialog and no double-tap seeks.
Proposed partial credit
30% paused near target (±5 s); 70% within ±1 s.
R01.11.02.01-T10 AntennaPod: seek to 12:34 and set playback speed 1.35x
Original instruction
In AntennaPod, play episode 'Ep 41 – Soil' and use the seek bar to reach 12:34 (±2 s), then set playback speed to 1.35× with the speed slider.
- Platforms
and· Android- Applications
- AntennaPod
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
AntennaPod DB/prefs: current episode position 754±2 s, playback speed 1.35 exactly; media session playing or paused. adb shell (media-key injection, am broadcasts, run-as DB/prefs edits) is not available to the agent, so position and speed must be set with the in-app seek bar and speed slider.
Proposed partial credit
50% position within ±2 s; 50% speed exact.
R01.11.02.01-T11 iOS: set Text Size to the third-largest standard step
Original instruction
On the iOS 18 Simulator, in Settings › Display & Brightness › Text Size, drag the slider to the third-largest standard step (not Larger Accessibility Sizes) and turn Bold Text on.
- Platforms
ios· iOS/iPadOS- Applications
- iOS Settings
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G2· GUI-essential- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
Appium mobile: getContentSize returns 'extra-large' (third-largest standard category) and bold text enabled in the simulator's accessibility prefs. Host shell and xcrun simctl are withheld from the agent, so 'simctl ui content_size' is unavailable; size and bold must be set in Settings.
Proposed partial credit
70% correct size category; 30% bold text on.
R01.11.02.01-T12 Android: trim a clip to 3.20–7.80 s with editor handles
Original instruction
In Open Video Editor, open Movies/pour.mp4 (30 fps) and drag the trim handles to keep 00:03.20–00:07.80, then export to Movies/pour_trim.mp4 without re-cropping or speed change.
- Platforms
and· Android- Applications
- Open Video Editor
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
touch· Touch gesturesdrag· Drag & dropav· Audio/video over time- Work product
WP05· Time-based media- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
ffprobe duration 4.60±0.07 s; first and last frames match source frames 96 and 234 (±2 frames) by SSIM ≥ 0.95; resolution and fps unchanged. adb and root shell are withheld from the agent (screen input only), so host file pushes are unavailable; the clip must be trimmed and exported in the editor.
Proposed partial credit
20% export exists; 40% start within ±2 frames; 40% end within ±2 frames.
R01.11.02.01-T13 Shop replica: dual-thumb price filter to $35–$120
Original instruction
In the ShopDemo app, on Headphones, set the price range slider (dual thumbs, $1 steps, no text inputs) to exactly $35–$120, add the 4.5+ rating filter, and add the cheapest remaining item to the cart.
- Platforms
and· Android- Applications
- ShopDemo
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
ShopDemo backend logs filter min=35 max=120 and rating≥4.5 at the time of add-to-cart; cart contains exactly the ground-truth SKU. adb is withheld and the ShopDemo backend accepts filter/cart calls only with the app's session token, so the price slider must be used.
Proposed partial credit
25% min thumb exact; 25% max thumb exact; 20% rating filter; 30% correct item in cart.
R01.11.02.01-T14 Thermostat replica: circular slider to 21.5 °C and schedule
Original instruction
In HomeDemo, set the living-room thermostat's circular dial to 21.5 °C (0.5 steps, no buttons), then open Schedule and drag the 'Night' block's start handle to 22:30.
- Platforms
and· Android- Applications
- HomeDemo
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturespath· Continuous pathdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
HomeDemo backend state: setpoint 21.5, schedule Night start 22:30 (15-min snap); other blocks unchanged. No adb shell or emulator console is exposed, and the HomeDemo backend accepts setpoint and schedule writes only from the app session (no API token issued).
Proposed partial credit
50% setpoint exact; 40% night start exact; 10% other blocks unchanged.
R01.11.02.01-T15 Android: fling a long snap picker to item 147 'Zr-87'
Original instruction
In LabPicker (test app), choose isotope 'Zr-87' from a snapping wheel of 300 alphabetic items (no search, no letter index) and confirm. The wheel decelerates with momentum.
- Platforms
and· Android- Applications
- LabPicker
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesnav· Pan/zoom/scroll- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 3 / 5
Proposed verifier
LabPicker result file shows selection index 146 ('Zr-87'); the app logs fling count for analysis. adb and root shell are withheld from the agent (screen input only), so am start extras and run-as result-file edits are unavailable; the picker must be flung.
Proposed partial credit
30% landed within ±3 items of 147; 70% exact item 147 'Zr-87' selected.
R01.11.02.01-T16 iOS Photos: exposure +0.35 and warmth −20 with ruler dials
Original instruction
In Photos on the iOS 18 Simulator, edit photo IMG_0021: set Exposure to +35 and Warmth to −20 using the horizontal ruler dials (values shown above the dial), then tap Done.
- Platforms
ios· iOS/iPadOS- Applications
- Photos
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & drop- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
Grader reads the asset's adjustment data from Photos.sqlite/resources: exposure 0.35 and warmth −0.20 (±0.01 normalised), no other adjustments. Host shell and xcrun simctl are withheld from the agent, so Photos.sqlite cannot be edited; adjustments must be made with the ruler dials.
Proposed partial credit
45% exposure exact; 45% warmth exact; 10% nothing else changed.
R01.11.02.01-T17 Android VLC: set an A-B loop around a spoken code word
Original instruction
Movies/briefing.mp4 contains the spoken word 'Tangerine' once. In VLC, find it by listening/scrubbing and set the A-B repeat loop starting ≤0.5 s before the word and ending ≤0.5 s after it.
- Platforms
and· Android- Applications
- VLC
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropav· Audio/video over time- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 5 / 5
Proposed verifier
VLC A-B loop points (from player state dump) satisfy A∈[t0−0.5, t0] and B∈[t1, t1+0.5] where t0/t1 come from forced alignment of the word. adb is withheld and VLC's Remote access server is disabled, so the A-B points must be set in the player UI.
Proposed partial credit
30% loop set around correct region (±3 s); 35% A tolerance; 35% B tolerance.
R01.11.02.01-T18 Android drawing app: match a target colour with hue ring and SV square
Original instruction
In Sketchbook-lite (test app), the hex field is disabled. Using only the hue ring and saturation/value square, set the brush colour to match the swatch #3A7BD5 shown on screen within ΔE2000 ≤ 2, then paint one stroke and save.
- Platforms
and· Android- Applications
- Sketchbook-lite
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropvis· Visual judgment- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 5 / 5
Proposed verifier
App prefs brush colour vs #3A7BD5: ΔE2000 ≤ 2 for full credit, ≤ 5 partial; saved image contains a stroke in that colour. adb and root shell are withheld from the agent (screen input only), so run-as pref edits are unavailable; the colour must be picked with the hue ring and SV square.
Proposed partial credit
30% hue within 5°; 50% ΔE2000 ≤ 2 (20 pts if ≤ 5); 20% stroke saved.
R01.11.02.01-T19 Mixer replica: four faders and two pan knobs, then save scene
Original instruction
In MixDemo, set channel faders to Vox −6 dB, Gtr −12 dB, Keys 0 dB, Click −∞ (fully down), pan Gtr L30 and Keys R30 (knobs, vertical drag), then save the scene as 'Rehearsal'. Faders snap to 0.5 dB.
- Platforms
and· Android- Applications
- MixDemo
- Interfaces
mob· Mobile touch- GUI need · task setting
G3· GUI-only- GUI need · ordinary access
G3· GUI-only- Interaction primitives
touch· Touch gesturesdrag· Drag & dropgeom· Precise geometry- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 4 / 5
Proposed verifier
MixDemo scene file 'Rehearsal': fader values exact at 0.5 dB resolution (±0.5 partial), Click muted at −inf, pans exactly L30/R30. adb and root shell are withheld from the agent (screen input only), so run-as scene-file edits are unavailable; faders and knobs must be dragged in the app.
Proposed partial credit
60% four faders exact (15 each); 30% two pans exact (15 each); 10% scene saved with name.
R01.11.02.01-T20 Control pair of T01: set 40% brightness via adb
Original instruction
Control pair of T01. With adb shell on the same AVD, set adaptive brightness off and screen brightness to 40% of range using settings commands, accounting for the brightness curve so the result matches what the slider would show.
- Platforms
and· Androidlin· Linux- Applications
- adb
- Interfaces
term· Terminal- GUI need · task setting
CL· CLI-sufficient- GUI need · ordinary access
CL· CLI-sufficient- Interaction primitives
code· Code & commands- Work product
WP08· App/system state- Conditions
- None declared
- Proposed difficulty
- 1 / 5
Proposed verifier
Same check as T01 (screen_brightness_float in [0.35, 0.45], mode 0); reported as the CLI baseline for the slider task.
Proposed partial credit
70% value in range; 30% adaptive off.
Mapped research references
Existing benchmark-to-class links from the research catalog. They are not record-level annotations, executions or product endorsements.