It's free.

No paid plans. The engine is open source under AGPL-3.0. You pay only your own LLM provider — or nothing, with a local model.

AGPL-3.0

Run it yourself

MonkeyTest Core is the CLI and library. Free forever. Install it, point it at a URL, and run it as often as you like. A hosted LLM bills you for tokens. Ollama, or another local model, does not.

  • Unlimited runs on your machine or CI
  • CLI: run, rerun, diff
  • Your LLM key, or a local model via Ollama
  • Screenshots, video, console and network capture
  • plan.json and run.json you can commit
  • Fork it, embed it, self-host it
$ npm install -g @monkeytestai/monkeytester
$ monkeytest run https://example.com

FAQ

Answers about the open-source engine and the live demo.

What is MonkeyTest.ai?

MonkeyTest is an open-source AI agent that uses your website like a real visitor — clicking, navigating, filling forms — and reports bugs with screenshots and video. Regular tests only check what you expected. monkeytest explores and finds what they miss.

Is the CLI really free?

Yes. MonkeyTest Core is the same engine, published on GitHub under AGPL-3.0. Install it from npm, point it at any URL, run it as often as you like — there is no paid edition, no seat limit and no run cap. You supply your own LLM API key, which for a small site typically costs a few cents per run. A local model via Ollama costs nothing extra.

What's the difference between the CLI and the live demo?

The CLI runs on your machine or your CI runner: unlimited runs, your own LLM or a local model, results as local files. The live demo at app.monkeytest.ai lets you watch a run with no setup. The demo includes 2 Scout attempts per day. Run it yourself for unlimited use.

Do I need to write test scripts?

No. There are no test scripts, fixtures, or selectors to maintain. Run the open-source CLI, or open the live demo and watch a run.

What kinds of bugs does it find?

Broken user flows, forms that silently fail, buttons that don't respond, navigation that dead-ends, JavaScript errors, missing or broken links, layout and responsiveness issues, and many common UX problems. It's not a replacement for deterministic critical-path tests, a load tester, or a security scanner — pair it with those for full coverage.

What is a run?

One run is one exploration of a URL: the agent looks at the site, tries the flows a real user would, and writes a report. On your own machine there is no cap. The live demo includes 2 Scout attempts per day.

Can I test a staging site or something behind auth?

Yes. Any publicly reachable URL works, including staging. Authenticated areas need credentials configured for the CLI — the docs cover that. Localhost works through a tunnel such as ngrok.

How does re-run work?

Every run produces a stable test plan. After you ship a fix, rerun executes that same plan again and a diff shows what is new, fixed, regressed, or still broken.

Can I share a bug report?

From the CLI, the report is files on your machine — screenshots plus plan.json and run.json — so you can commit them or send them to whoever needs them.

Are there paid plans?

No. The engine is free under AGPL-3.0. You pay your LLM provider only if you use one, or nothing with a local model. The live demo is free and includes 2 Scout attempts per day.