QA you never write or maintain.

Describe what should work in plain English. Monito proves it in a real browser on every deploy and on a schedule, with screenshots, network and console evidence behind every verdict. When we're the one that fails, you don't pay.

Run your first test — it's freeSee how it works

Animated demo of a coding agent calling Monito's run_scenario MCP tool: the Monito agent tests acme.dev/checkout in a real browser, submits invalid form data, captures screenshots, finds a crash on negative quantity and reports one bug, which the coding agent then reads with get_run.

0
Fig. 1 false passes in 111 runs on our test corpus
0 lines
Fig. 2 of test code to write or maintain
$0
Fig. 3 when Monito is the one that fails

Describe it once. Monito keeps proving it.

01

Describe

Write what should work, in plain English: A new user can sign up and lands on an empty dashboard. Not sure where to start? Discovery explores your app and proposes tests to review.

prompt:"A new user can sign up and lands on an empty dashboard."
02

Prove

Monito does it in a real browser and returns passed, failed or cannot verify. A pass has to cite a measured check. If Monito can't prove it, it doesn't pass.

navigate/signup
fillemail
click'Sign up'
03

Keep proving

On Pro, run it on every deploy (GitHub Actions, a deploy webhook or Vercel previews) or on a schedule. You hear from Monito when something breaks.

FAILSignup failed
Run your first test free

A pass you can check.

AI testing tools ask you to trust them. Monito shows its work.

A pass has to be proven

Every expected behavior must cite a measured check: an assertion on the page or a visual check of a screenshot. No proof, no pass.

It never changes what you expect

Monito never rewrites a test to make it pass. When your app changes, you get a finding, not a quietly healed test.

Accuracy we publish

Right 110 of 111 times on our test corpus, with 0 false passes and 0 false fails. Every run is public, and so are the limits. How we measure

Our mistakes are free

If Monito caused the error, the run costs nothing and its credits come back. A test that finds a bug in your app still counts.

Reports anyone can open

Every run is a report your team can read: steps, screenshots, network and console, and the exact checks behind the verdict.

Your journeys on a status page

Publish the flows that matter on a public status page, verified in a real browser (on Pro). We test ourselves the same way: sign in, create a project, run a test, read the docs.

See ours

Works where you build.

On your machine, in CI and from your coding agent.

  • Check localhost

    monito check --local 3000 "…" tests your dev server through a private tunnel from your machine. Nothing listening, no run, no charge.

  • A verdict for your coding agent

    Over MCP or the CLI, your agent gets the verdict and the evidence back. The browser stays out of its context, and your saved login passwords never reach the model.

  • Stop a bad deploy

    A failed run exits non-zero, so CI stops the build. On Pro, the GitHub Actions integration posts one check per run and updates a single PR comment when needed.

  • Evidence, not a summary

    Bug reports come with repro steps, failed network requests, console errors and screenshots. Add --json to any CLI command for output you can pipe.

For scripts, CI and any agent that runs shell commands. Every command takes --json, and a failed run exits non-zero.

$npm i -g @monitodev/cli
Read the CLI docsView on npm
monito cli@monitodev/cli
$npm i -g @monitodev/cli
+ @monitodev/cli
$monito project run web-checkout
✓queued run_8f2a1c
navigate/checkout
fillcard → 4242 4242 4242 4242
bugcrash on negative quantity
FAIL1:24 · 47 events · 1 bug · 6 screenshots
$echo $?# non-zero → stops the deploy
2
Hosted MCP server
https://www.monito.dev/api/mcp

Nothing to install. Add the URL to your client and sign in to Monito in the browser when it asks.

$claude mcp add --transport http monito https://www.monito.dev/api/mcp
Read the MCP docsRead the CLI docs
coding agentmonito mcp
>Add a quantity field to checkout and check it in the browser
callrun_scenario · Checkout quantity
Run run_8f2a1c failed: 1 bug, 84s. Report: monito.dev/session/8f2a1c
FAIL1:24 · 1 bug
callget_run · run_8f2a1c
bugCart crashes when quantity is -1
1. Open /checkout 2. Set quantity to -1 3. Click “Update cart”
console: TypeError: Cannot read properties of undefined (reading 'price')
editsrc/cart/quantity.ts · clamp quantity to 1–99
callrun_scenario · Checkout quantity
PASS1:02 · 0 bugs

Frequently asked questions

A pass must cite a measured check for each expected behavior; without one the run doesn't pass. On our test corpus it was right 110 of 111 times with 0 false passes. The corpus is small and synthetic, so we publish every run and its limits. See every run

Start free. Pay for what runs.

One credit is one test run. A discovery costs 20. No seats, no per-project fees.

Free
$0

50 free test credits to start, no card

  • Manual runs and Discovery included
  • Run evidence kept 30 days
Run your first test
ProPopular
$79/mo

For founders and small teams shipping every week

700 credits each month

≈ 600 tests + 5 discoveries

  • Unlimited projects and environments
  • Automations: schedules, deploy webhook, GitHub Actions
  • Public status page
  • API keys
  • Bug reports with screenshots, network and console evidence
  • Run evidence kept 90 days
Team
$199/mo

For teams running suites on every deploy

1,800 credits each month

≈ 1,500 tests + 15 discoveries

  • Everything in Pro
  • Slack notifications
  • Run evidence kept 180 days
  • Priority support
Enterprise
Custom

For larger teams with their own limits and requirements

  • Custom credit volume
  • Custom data retention
  • Direct line to the founder
Talk to us

50 free test credits to start, no card. Without a plan, run evidence is kept 30 days. If Monito caused the error, you don't pay.