Agentic Exploration

Discover your app's flows
continuously

Paste your URL. Kane explores your running app, shows you what is broken and designs the test cases to prove it works.

Get KaneAI Max free for a month. Worth $179.

No PRD or test scripts needed

Explore the whole application

Kane reaches every screen and connects them on one map.

Findings show what needs fixing

Functional, backend and accessibility issues, each with evidence.

Assurance designs the test cases

Use cases, acceptance criteria and test cases for end-to-end coverage.

What is Kane Explore

Kane Explore is how Kane learns your app. Parallel browser agents explore every screen and user flow, report findings with evidence and hand the map to Kane Assurance for end-to-end coverage.

How it works

From one URL
to real test cases.

Browser agents explore different screens and user flows in parallel, logging every request and console error. No scripts to write.

App URL northstar.example.com20 screens explored
agent 01/
agent 02/products
agent 03/products/tote
agent 04/cart
agent 05/checkout
agent 06/login
agent 07/signup
agent 08/account
agent 09/orders
agent 10/orders/1042
agent 11/search
agent 12/category/bags
agent 13/wishlist
agent 14/addresses
agent 15/payment
agent 16/returns
agent 17/contact
agent 18/help
agent 19/profile
agent 20/reset
Use cases

Know what works
before your users do.

Verify what your AI agent built

Your coding agent says the change works. Run Kane on the preview URL and catch broken flows, dead buttons and failed requests before code review starts.

Catch regressions before release

A hotfix or a big release just reached staging. Kane explores it again and shows new screens, changed flows and the findings that are back.

Understand an app you inherited

No docs, no tests and nobody left who built it. Give Kane the URL and get every screen, every user flow and the PRD the app never had.

App map20 screens · 12 flows
HomeSearchProductCartCheckoutAccountOrders
Get started

Start from your browser
or your terminal.

KaneAI

Explore from your browser

Paste a staging or preview URL. The run is read-only, so your data stays untouched.

Explore your app
Kane CLI

Run from your terminal or with your coding agent

One command from your terminal, or one prompt for your coding agent.

$ npm install -g @testmuai/kane-cli
$ npx @testmuai/kane-cli-skill
$ kane-cli login
$ kane-cli explore run --project <id> \
    --url https://your-app.example.com --policy read_only
$ kane-cli cover
Works withClaude CodeCodexCursorand other coding agents
FAQ

Your questions answered

How Kane explores, what it checks and how it treats your application.

What is Kane Explore?

Kane Explore is the agentic exploration capability of Kane, the testing agent from TestMu AI (formerly LambdaTest). Starting from a single URL, parallel browser agents explore your running application, map its screens and flows, and report issues with browser evidence. You write no test scripts.

How much of my app will Kane cover?

Every screen it can reach within the run limits. Kane stops at a login it has no account for, so add a test account for the screens behind sign-in. It covers breadth across your app, and a person decides which flows need deeper tests.

Is every finding a real bug?

No. A finding is a lead with evidence attached. It stays open until a person reviews it and decides whether it is a real bug. Kane does not know your business rules, such as which discount a customer should get, so it cannot judge those on its own.

Does Kane modify the application while exploring?

Not from this page. Runs started here are read-only, so Kane blocks POST, PUT, PATCH and DELETE requests. Kane CLI allows changes by default, so pass --policy read_only for the same behaviour. Blocking those requests does not guarantee every interaction is free of side effects, so review the policy before exploring a sensitive environment.

How does Kane explore a web application?

You give Kane the URL of your running web application. For an app behind login you can add a test account. Kane's browser agents follow links, controls and forms in parallel, discover screens and map the flows that connect them.

What are parallel browser agents?

They are separate cloud browser sessions that Kane runs at the same time. Each agent explores a different screen or flow, so a multi-screen application is mapped faster. Kane Explore supports up to 20 browsers in parallel.

Can Kane discover application issues automatically?

Yes. Kane checks discovered screens for console errors, failed network requests, broken links, slow requests and accessibility issues. You do not write assertions first.

Does Kane provide evidence for its findings?

Yes. Each finding links to the screen and the flow where Kane saw it. It includes browser evidence, such as the failed request or the console error. Your team has a concrete place to start.

How does Kane turn exploration into test design?

Discovered screens and flows become application context. From that context Kane derives business entities, use cases, acceptance criteria and linked test cases. Exploration designs tests and does not execute them. A person reviews the test design before choosing to run it.

My coding agent already drives a browser. Why use Kane?

Your coding agent checks the change it just made. Kane explores the whole running app and is independent of the agent that wrote the code. It also flags issues nobody asked it to look for.

I already have a Playwright suite. Why explore?

A scripted suite checks the flows someone wrote assertions for. Exploration maps the screens and flows the suite never touched and flags issues nobody asked it to look for. The two work together.

Can I run Kane from Claude Code, Cursor or Codex?

Yes. Install @testmuai/kane-cli and give your coding agent the Kane CLI commands. After a change it can run kane-cli explore run and kane-cli cover, then summarise the application map and the findings.

How long does an exploration run?

A run ends when the agents have nothing left to explore or when it reaches a built-in limit: one hour, 400 actions or 800 model calls. From Kane CLI you can also set your own cap.

Does it work on mobile apps?

Not yet. Kane Explore works on web applications today.

Your app is built.
Now find out what it does.

Paste one URL. Kane shows you the map, the findings and the test cases to review.

Explore your app