Effortless end-to-end testing

Record tests while you build your app, let your AI coding agent write and run them, and run every test on Chrome, Edge and Firefox in parallel with a single line in CI.

No test code to write
Built for AI coding agents
Parallel cross-browser runs

Develop and test the UI at the same time

Open the Mimicus side panel next to your app on localhost and keep working. Every click and every value you type becomes a test step, and you add checks by pointing at an element. When you're done with the feature, its test is already done too.

Replay instantly after each code change, in the same tab
Chain scenarios: start from a signed-in state instead of re-recording the login
Robust locators built from labels, text and test ids, not brittle CSS paths

Recording · Checkout: pay with card

Navigate/cart
Clickbutton “Checkout”
Fill“Card number” ← 4242 4242 4242 4242
Clickbutton “Pay now”
Expecttext “Thanks for your order” is visible

Made for AI end-to-end testing

Coding agents are bad at checking UI work by driving a browser and reading screenshots: it's slow, flaky and burns tokens. With Mimicus the agent writes a small JSON scenario and runs one command. It gets a clear pass or fail and a short, readable reason when something breaks.

One command starts your app, runs the tests and shuts it down
Exit code 0 means it works, so agents verify their own changes
Ready-made agent skill teaches your agent the scenario format

// scenarios/todos.json, written by your agent
{
  "name": "Todos: add an item",
  "steps": [
    { "type": "Navigate", "url": "/todos" },
    { "type": "Fill", "locator": { "type": "Css", "selector": "input[name=title]" }, "value": "Buy milk" },
    { "type": "Click", "locator": { "type": "Text", "text": "Add" } }
  ],
  "expect": [{ "assertionType": "ElementVisible", "locator": { "type": "Text", "text": "Buy milk" } }]
}

$ mimic run -f scenarios --base-url http://localhost:5173 --start "npm start"
✓ Todos: add an item  1.2s

One line in CI, every browser in parallel

The tests you recorded or your agent wrote run in CI without any extra setup. One command sends them to Mimicus cloud workers, which run them in parallel on real browsers and report results straight back to your pipeline.

No browser setup: no grid to host, no drivers to keep in sync
Parallel by default, so the suite stays fast as it grows
Run a subset with groups like -g smoke

ChromeEdgeFirefox

# .github/workflows/e2e.yml
- run: npx @tonys-software/mimic run --cloud

Started run 7f3c…: 24 scenarios on the cloud workers
✓ Checkout: pay with card    chrome   2.1s
✓ Checkout: pay with card    firefox  2.6s
✓ Checkout: pay with card    edge     2.2s
…
24 passed