---
title: "Kane CLI launches today: a free terminal tool that runs the browser check AI agents can't"
description: "TestMu (formerly LambdaTest) shipped Kane CLI today: a free terminal tool that runs the browser check AI coding agents like Claude Code cannot do."
author: Darie Nani (Editor-in-Chief)
date: 2026-04-28T13:11:18.485Z
updated: 2026-04-28T13:11:18.497Z
canonical: https://www.sovereignmagazine.com/article/testmu-kane-cli-launch
image: https://cdn.nanimediahouse.com/kane-cli-featured.webp
categories: Artificial Intelligence, Startups
content_type: News, Spotlight
region: San Francisco
publication: Sovereign Magazine
about:
  - type: Organization
    name: TestMu AI
    description: TestMu AI (Formerly LambdaTest) is a fully autonomous agentic quality engineering platform that empowers teams to test intelligently and ship faster. Built for scale, it offers a full-stack testing cloud with AI Agents for planning, authoring, executing, and analyzing software quality at scale. The platform can be used to test any type of software app, including web, mobile, and enterprise applications, at any scale, and in any type of environment, including real devices and browsers. Find out more at www.testmu.ai
    url: https://www.testmu.ai/
    sameAs:
      - https://www.facebook.com/lambdatest/, https://twitter.com/Lambdatesting, https://www.linkedin.com/company/lambdatest/, https://www.youtube.com/channel/UCCymWVaTozpEng_ep0mdUyw?sub_confirmation=1, https://github.com/LambdaTest/, https://www.pinterest.com/lambdatest/
---

TestMu AI today launched Kane CLI, a browser automation tool that runs from a developer's terminal. Users install it via npm or Homebrew, describe a flow in plain English (log in, add an item to cart, check out), and Kane CLI opens a real Chrome window to run through the steps and report whether each one worked.

The launch arrives at a particular moment for AI coding tools. Over the past 18 months, [Claude Code](https://www.sovereignmagazine.com/article/claude-code-security-is-what-cowork-should-have-been), Cursor, Codex CLI and Gemini CLI have made it possible to write working software from a written prompt. None of these tools, however, can open a browser and confirm that the code they have just produced behaves the way the prompt intended. That step still requires a human.

Asad Khan, co-founder and chief executive of [TestMu AI](https://www.testmuai.com/) (the company most readers will know by its previous name, LambdaTest), argues that this gap is the next significant bottleneck for software teams. "Every feature that ships from a prompt is a feature nobody has actually verified," Khan said in a statement accompanying the launch. "At agentic speed, 'a human will click through it later' is not a plan. It's a liability."

> "Every feature that ships from a prompt is a feature nobody has actually verified. At agentic speed, 'a human will click through it later' is not a plan."
> — Asad Khan, CEO, TestMu AI

## How Kane CLI works: terminal-native browser automation in three modes

Kane CLI ships in three modes. The first is a terminal interface for developers to use by hand, where a flow can be typed in, run, watched and adjusted on screen. The second is a one-shot command that runs without a visible browser, intended for shell scripts and continuous integration pipelines. The third mode is built for AI coding agents to call directly: the agent passes the tool a flow and receives a structured result it can act on. Claude Code, Codex CLI, Cursor and Gemini CLI all read this output natively.

Several features distinguish Kane CLI from the Playwright and Selenium scripts most engineering teams write today.

Tests are described in plain English. Developers do not have to write CSS selectors, XPath expressions or other element identifiers, which break whenever a designer renames a class or restructures a page.

The tool persists when something on the page changes. Kane CLI attempts up to fifty steps to complete a flow before giving up, where most existing tools fail at the first unexpected element.

When Kane CLI hits a one-time password prompt or a CAPTCHA, it pauses and asks the human to handle that step before continuing. For AI coding agents running automated tests overnight, this prevents a single failed step from abandoning the entire flow.

The tool is two-way compatible with Playwright. Existing Playwright tests can be converted to Kane CLI and vice versa. TestMu has positioned this deliberately to lower the cost of switching: teams currently using LambdaTest, Sauce Labs or BrowserStack do not need to discard their existing test suites.

## Why AI-written code needs a browser check before it ships

Khan describes the problem the tool addresses as a trust problem rather than a testing problem. "Software has always trusted the people who wrote it," he said. "Now, for the first time, it has to trust the machines."

Engineering teams at several large software companies have published similar observations over the past year. Render reported in late 2025 that its engineers are now opening multiple pull requests a day instead of one or two a week, and that the bottleneck in their development process has shifted from writing code to reviewing it. Meta's engineering team published an article in February 2026 titled "The Death of Traditional Testing", arguing that the QA model that has held for the last fifty years no longer fits how AI-augmented teams build software.

Demand for a way to verify AI-generated code is therefore real. Whether Kane CLI is the right answer is not yet clear. TestMu's existing scale and customer base, however, give it a credible starting position.

## Why LambdaTest rebranded to TestMu AI in 2026

Most readers will know the company by its previous name, [LambdaTest](https://www.sovereignmagazine.com/article/who-tests-the-bots-inside-lambdatest-s-agent-to-agent-qa-experiment). It operated a cloud-based Selenium grid that allowed development teams to run browser tests across thousands of browser and operating system combinations without provisioning their own hardware. The company was founded in 2017 by Khan and co-founder Jay Singh, with offices in India and the United States. By 2024 it reported approximately $120 million in annual revenue, more than 18,000 enterprise customers and a total of $108 million raised across eight funding rounds.

In December 2024, Avataar Ventures led a $38 million Series D round at a $400 million valuation, with Qualcomm Ventures participating. The case made to investors was that traditional scripted testing would not survive the volume of AI-generated code being produced, and that whichever company solved the resulting verification problem would command a higher valuation than a pure Selenium grid operator.

In January 2026, the company rebranded to TestMu AI. The "TestMu" name comes from its annual community conference. The new positioning describes the company as an "[agentic quality engineering](https://www.sovereignmagazine.com/article/testmu-ai-agents-will-vibe-test-your-vibe-code)" platform, a category label that Tricentis, Perfecto and a small number of other testing vendors have also adopted. Industry analyst firm Gartner has not yet formally recognized this category, and whether it endures remains an open question.

## TestMu vs Browserbase: the AI browser automation market

The most direct competitor to Kane CLI is Browserbase, a Y Combinator-backed company founded in January 2024 by Paul Klein IV. Browserbase raised a $40 million Series B round in June 2025 at a $300 million valuation, with Notable Capital and Kleiner Perkins among its institutional backers and angel investments from Stripe co-founder Patrick Collison, Vercel chief executive Guillermo Rauch and Twilio founder Jeff Lawson. The company has reported approximately $67 million in annual recurring revenue. Browserbase sells cloud-hosted browsers for AI agents to use, and its open-source SDK, Stagehand, has been widely adopted by developers building browser-driving AI agents.

The two companies are targeting the same end buyers (software engineering teams using AI coding agents) but with very different product shapes. Browserbase is sold as cloud infrastructure with an SDK layer above it. Kane CLI is a free local download that integrates with whichever coding agent the developer is already using.

Beyond these two, Microsoft's open-source Playwright library remains the underlying technology that Kane CLI imports from and exports to. Anthropic's Computer Use API and OpenAI's ChatGPT Agent (which absorbed the original Operator product in mid-2025) are also able to control a browser, but both are general-purpose tools rather than dedicated testing products. Among companies building browser-testing tools specifically for AI coding agents, TestMu and Browserbase are currently the only two with significant venture backing and a public product.

## Kane CLI pricing, free tier and what to watch next

Kane CLI is free to install. TestMu offers a paid tier above the free product, which adds cloud test runs, parallel browser sessions and team management features. Customers signing up during the introductory period receive three months of bonus credits. The free entry point is commercially significant: by placing Kane CLI inside the same terminals that already run Claude Code and Cursor, TestMu has positioned the paid product as the next purchase for teams that adopt the free version.

Whether engineering teams take that step is not yet clear. Software engineers have been wary of automated-testing promises before, having seen previous categories named and abandoned. Kane CLI will be useful if it removes manual clicking. It will damage trust if its results turn out to be unreliable. The next three months of real-world use should produce more evidence on which of those is happening.

For developers and AI coding agents that want a verification step before pushing code, Kane CLI is now an option. The product is free to install today.

**About TestMu AI**

TestMu AI (Formerly LambdaTest) is a fully autonomous agentic quality engineering platform that empowers teams to test intelligently and ship faster. Built for scale, it offers a full-stack testing cloud with AI Agents for planning, authoring, executing, and analyzing software quality at scale. The platform can be used to test any type of software app, including web, mobile, and enterprise applications, at any scale, and in any type of environment, including real devices and browsers. Find out more at www.testmu.ai

[Website](https://www.testmu.ai/)

## FAQ

**Q: What is agentic testing?**
Agentic testing is a label for a new generation of testing tools that act on intent rather than scripts. Instead of telling a tool to click a specific button by its CSS selector, you describe the goal in plain English and the tool figures out how to do it. Kane CLI is one example. Tricentis, Perfecto and a few others are using the same label for similar products.

**Q: Can AI do QA testing?**
Yes, with caveats. AI tools can now describe a test, run it in a real browser, and report what worked and what did not. They are good at flows that change often, where traditional scripts break. They are still weak at finding bugs that need product judgment rather than mechanical checking. For visual regression and end-to-end flows, AI testing is reliable enough to use in production today. For deeper quality work, a human is still needed in the loop.

**Q: What is Kane CLI from TestMu AI?**
Kane CLI is a terminal-native browser automation tool that launched on April 28, 2026. It is free to start, installs via npm or Homebrew, and works with Claude Code, Codex CLI, Cursor and Gemini CLI. Developers and AI agents use it to verify whether a website behaves the way they expect. It also exports tests to Playwright and runs in CI pipelines like GitHub Actions.

**Q: Which AI is best for coding tests?**
It depends on what you mean. For writing the application code, Claude Code, Cursor, Codex CLI and Gemini CLI are the four most widely used coding agents in 2026. For checking that the code those agents produced actually works in a browser, Kane CLI is the first tool built specifically for that step. For unit and integration tests inside the codebase, the coding agents themselves usually generate them as part of writing the feature.