Evidence,
not guesswork.
One precision platform for every signal your browser can observe: controllers, mice, keyboards, displays, audio, cameras and system hardware.
Choose the signal. The test is already ready.
Every module opens directly into a focused test. No account, no setup wizard and no raw device stream sent to InputProof.
Controller
Stick drift, idle stability, live axes, buttons and an evidence report.
↗Mouse
Movement rate, five buttons, holds, click timing and wheel deltas.
↗Keyboard
Keydown, keyup, repeats, physical codes and concurrent holds.
↗Display
Geometry, estimated refresh, motion reference and pixel color sweep.
↗Audio
Microphone level, local device paths and a speaker reference tone.
↗Camera
Private preview, negotiated resolution, frame rate and local still.
↗Browser hardware
CPU, memory, GPU, screen, storage, network and platform APIs.
↗Thirty-seven focused questions. Seven shared engines.
Start from the exact symptom, inspect the method and boundary, then open the live protocol at the relevant module.
Browse all 37 tests ↗Stick drift
Capture a five-second idle window and separate persistent center offset from isolated stick noise.
↗02MOVE / 5 BUTTONS / WHEELEvent rate
Run a fixed eight-second pointer capture with Low 5%, jitter, stability and dropout evidence.
↗03KEY / CODE / LOCATIONEvery key
Verify physical codes, labels, locations, down/up state and full-layout coverage.
↗04PIXELS / FRAME TIME / COLORRefresh delivery
Measure a fixed eight-second requestAnimationFrame window with cadence and confidence evidence.
↗05PCM LEVEL / DEVICE PATHMicrophone
Inspect live RMS dBFS, peak, noise floor, waveform, FFT and processing flags.
↗06VIDEO STREAM / SETTINGSWebcam
Open a private local preview and inspect negotiated resolution, FPS, aspect and controls.
↗07EXPOSED SYSTEM CAPABILITIESWebGL
Inspect WebGL context attributes, limits and exposed extensions without rendering a stable fingerprint.
↗Your controller, under evidence.
- WINDOW
- 5.0 SEC
- PRIMARY
- RADIAL P95
- HEALTHY
- ≤ 0.050
- MINOR
- 0.051–0.100
- SIGNIFICANT
- > 0.100
- FRAME QUALITY
- AWAITING
- STORAGE
- MEMORY ONLY
Trace both stick edges through full circles.
Move both sticks around the outer gate several times. Range, angular coverage and circularity are calculated from browser-mapped axes.
This previews an application-layer radial deadzone. It does not change controller firmware or game settings.
Connect a controller or load a reference trace
No active Gamepad API signal source is available.
Connect a controller or load a reference trace
No active Gamepad API signal source is available.
- The controller is disconnected or has not produced an input event.
- Connect the controller, press a button and keep this tab visible.
The active device name and live axes must appear before capture.
Actions are generated from visible Evidence fields and versioned product rules. They do not identify a physical root cause with laboratory certainty.
Compare repeat tests without an account.
Complete this protocol before saving a baseline.
Save 3 more compatible run(s).
A result should show its work.
Inputs, media and system capabilities use the same product rule: expose the signal, state the browser boundary and never turn “not observed” into “failed.”
Patterns over panic.
A single spike is not a diagnosis. We measure a timed foreground window and use robust statistics to reduce false alarms.
Certainty is earned.
Sample count, tab visibility, connection stability and test completion shape the confidence rating.
Repair before replacement.
Free fixes and confirmation come first. Commercial recommendations only follow repeatable evidence.
Broad coverage. Honest boundaries.
Seven active suites now cover input devices, visual output, media paths and browser-exposed hardware. Deeper protocols can grow without changing the one-action interaction model.
Stop wondering.
Measure the signal.
Open device lab No account. No install. No raw device signal uploaded.