Fuzz your UI.
Reproduce every bug.

The deterministic UI fuzzer for every platform.

Private beta/ CI-native/ Language-agnostic
evidence, repro@case-0042 SEED 0
Flutter React Native SwiftUI UIKit Jetpack Compose Web Electron Tauri WinUI WPF AppKit GTK Dear ImGui Clay TUI
ci / nightly · run 4,127 3 flaky
The problem

Right now, your users find the bugs first.

01

Every bug lands on a real person.

A dead button, a frozen screen, a checkout that silently fails. To a user that's not an edge case, it's frustration, a one-star review, and a reason to leave.

02

Tests only walk the paths you wrote.

Scripted tests cover the happy flow. The bugs that hit some users and not others live in the states, inputs, and tap orders you never thought to test.

03

And the worst ones won't reproduce.

Intermittent, device-specific, three taps deep. Without the exact seed, inputs, and state, you can't make it happen again, let alone fix it.

The solution

Fuzz it. Pin it. Replay it on demand.

AI maps your UI into a graph it understands. A deterministic fuzzer drives that graph until something breaks, then seals each failure as evidence you can replay forever.

01 · FUZZ

Fuzzer hunts

Guided by an AI map of your screens, it drives thousands of taps and swipes, chasing the path that breaks.
02 · CAPTURE

Seed & state pinned

The instant it breaks, the seed, inputs and environment freeze into a sealed evidence file.
seed 0x8f3a2c91
inputs · env · 3.2 KB
SEALED ✓
03 · REPLAY

Deterministic replay

Feed the seed to the runner and the exact same failure happens, 100 times out of 100.
App
item 1 / 3
Next →
!
App crashed
index 3 ∉ 0..2
reproduced 100 / 100 · 0.4s
checkout-agent #482 open
repro-itbot · pinned a non-deterministic failure

Found a checkout race that fails ~1 in 12k runs. I captured it. Reproduce the exact failure locally:

one-line reprocopy
$ repro run --seed 0x8f3a2c91
deterministic 100/100 runs 0.4s replay evidence 3.2 KB
In your pull request

A repro you can paste, not a trace you have to decode.

One command, zero setup.

Anyone on the team runs it and watches the same failure happen, no special branch, no “works on my machine.”

Blocks the merge until it's fixed.

The repro runs in CI. It stays red while the bug lives and flips green the instant your fix holds.

The evidence travels with the bug.

Seed, inputs and state are sealed in one portable file, attached to the PR, versioned with your code.

The kit

Built for systems that refuse to behave the same way twice.

Deterministic runner

Pins randomness, time, and ordering so a captured failure replays bit-for-bit, every single run.

core engine

Sealed evidence files

Seed, inputs, env and timing captured into one portable artifact you can commit, share, and re-run anywhere.

~3 KB · portable

Agent-aware fuzzing

Understands tool calls, model sampling and multi-step plans, surfaces the failures black-box testing never finds.

for AI agents

Time-travel to failure

Jump straight to the exact step the system broke. Inspect state, step backward, watch it fail in slow motion.

step debugger

CI-native, PR-first

Drops a one-line repro into the pull request and gates the merge. Works with the pipeline you already run.

GitHub · GitLab · CI

Language-agnostic

Wrap any binary, service, or agent runtime. If it has inputs and an exit code, Repro It can pin it.

any stack
Built for AI agents

Non-determinism is the default. Treat it like evidence, not fate.

Agents fail in ways unit tests were never built to catch, a tool called in the wrong order, a sampled token that cascades, a race three steps deep. Repro It pins the run so you can fix the cause, not chase the symptom.

1 / 12k
failure rate, caught and pinned
100%
replay fidelity from one seed
Pricing

Free to run yourself. Paid to host it.

The runner is free and source-available. The cloud is a paid, hosted service, usage-based on runs.

CLI

Free

Run it yourself, on your machine and in CI.

  • Source-available (Elastic License v2)
  • Web, Android and iOS
  • Deterministic fuzz + replay
  • No account needed
Get the CLI
Most popular

Pro

$49 / month

For shipping teams.

  • 5 projects, generous seats
  • 2,000 web + Android runs / mo
  • The built-in agent fix loop
  • 90-day retention, email support
  • Overage $0.02 / run
Coming soon

Team

$199 / month

More scale, plus mobile and SSO.

  • 12,000 runs / month
  • iOS device runs (metered)
  • Google SSO, 1-year retention
  • Priority support
Coming soon

Enterprise

Custom

For regulated teams.

  • SSO (SAML / OIDC) + SCIM
  • Self-host options
  • Dedicated device pool, SLAs
  • Security review & DPA
Contact us

The CLI is free and available now. Cloud plans launch shortly; annual billing saves ~2 months. iOS device runs roll out on Team and Enterprise.

Open source

Stop guessing.
Repro It

Install the CLI and pin your flakiest test or most haunted agent in minutes. Free and source-available.

Free, source-available CLI. The cloud is a paid hosted service.