Kane CLI
A CLI from TestMu AI that writes end-to-end tests from requirements and runs them in a browser or on mobile, with a skill for your agent
- 263 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Controls a real browser or device: clicks buttons and enters and submits form data
- After sign-in it uploads runs, steps and screenshots to the TestMu AI dashboard
- Natural-language tests act with your session's rights on the sites under test
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/kane-cliDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Controls a real browser or device: clicks buttons and enters and submits form data
- After sign-in it uploads runs, steps and screenshots to the TestMu AI dashboard
- Natural-language tests act with your session's rights on the sites under test
This is third-party code. Review the repository files before installing.
What it does
Kane CLI takes a goal in plain language, launches your installed Chrome through the DevTools protocol and executes steps in the defined scenario. It returns an NDJSON stream of steps and a result, and the exit code is usable in CI. The same tests run on desktop web, on Android and iOS emulators and simulators, and on real devices in the TestMu AI cloud. Test generation is handled by the assurance commands. kane-cli context ingest takes a requirements document, extract proposes use cases that cite their source, review leaves the decision to you, and design tests produces acceptance criteria, scenarios and one runnable _test.md file per scenario. A run saves a pack of screenshots, steps and logs, kane-cli cover shows what is proven per criterion, and maintain reconcile updates the suite when requirements change. The kane-cli skill for Cursor, Claude Code, GitHub Copilot, Codex and Gemini ships in the repository. It requires a ready check at the start of a session, runs tasks with the --agent flag, shows a result card and allows tests to be written only through the assurance commands.
Who it is for. For QA engineers and developers who want end-to-end tests built from requirements, and for anyone who wants Claude Code or Cursor to check its own code in a real browser.
Good fit when
- You have a PRD or specification and want scenarios, acceptance criteria and tests tied to the requirements
- After a change the agent must check sign-in, checkout or a form in a browser and return a structured result
- You want tests stored as Markdown in the repository and run in CI with kane-cli testrun run --tags smoke
- The app has to run on an Android emulator, an iOS simulator or a real device in the TestMu AI cloud
- You need a coverage report: what is proven per acceptance criterion and what is still owed
Not a fit when
- You need unit tests or library tests: Kane works at the level of user scenarios in a browser or app
- The machine has no stable-channel Google Chrome: the CLI launches the installed Chrome
- You need real devices without a plan: they require a TestMu AI plan
- You do not want run results on the TestMu AI dashboard: the CLI uploads the run there
- The agent should write tests by hand in chat: the skill forbids it and routes work through the assurance commands
Example request
Open https://news.ycombinator.com with kane-cli, extract the top story’s title using `store as top_story`, and show the resultLimitations
Installing through npm requires Node.js 18 or later. Desktop web tests need stable Chrome. Local Android emulators and iOS simulators run on Apple Silicon macOS and require Android Studio or Xcode. Real devices and remote suites depend on a TestMu AI cloud plan; remote execution also requires its plugin.
The CLI signs in through OAuth, uses billed service capabilities and sends run data to the cloud. Review test data, screenshots and network logs. Viewing a saved pack with `evidence serve` is local, but this does not apply to the whole CLI.
The skill describes readiness checks and an agent workflow. A passing selected scenario does not establish the absence of other defects. The repository uses Apache-2.0.
How to disable. First run `kane-cli logout` to revoke OAuth and remove the profile’s local credentials. Then uninstall the CLI using its installation method: for npm, run `npm uninstall -g @testmuai/kane-cli`. Remove the skill directory from the agent. Homebrew and script uninstall steps are documented in installation.md.
FAQ
How do I install Kane CLI?
With npm install -g @testmuai/kane-cli on Node 18 or newer, with brew install LambdaTest/kane/kane-cli through Homebrew, or with the install.sh script. Stable-channel Chrome must be present.
How do I connect Kane CLI to Claude Code or Cursor?
Run `npx @testmuai/kane-cli-skill` for a supported agent. Install the CLI separately. Skill copies also exist under .claude/skills and .agents/skills in the repository; run the readiness script from the installed skill directory.
How do I generate tests from requirements?
Load a document with kane-cli context ingest ./prd.md, then run context extract and context review. After that design tests --use-case uc-checkout creates acceptance criteria, scenarios and one _test.md file per scenario.
How do I run saved tests in CI?
A single file runs with kane-cli testmd run tests/checkout_test.md --agent, and a tagged suite with kane-cli testrun run --tags smoke --parallel 4 < /dev/null. The stdin redirect makes the output an NDJSON stream.
Where can tests run?
On desktop Chrome by default, on an Android emulator and iOS simulator with --target emulator and --target simulator, and on real devices with --target android and --target ios in the TestMu AI cloud under a plan.
Discussion
No questions or reviews yet.
Related
CLI toolsA skill that tests your agent's code in a real browser: reads your changes, builds a test plan and runs it through Playwright with your own cookies
MCP server, agent and skills for end-to-end UI tests where the agent drives the computer through vision on cloud machines
Seven QA agents and ten skills for test automation: Playwright in TypeScript, Selenium in Java, API, accessibility and ISTQB.
Guides for agents writing and fixing Playwright tests, with advice on locators, waiting and diagnosing CI failures
QA agents, skills and an MCP server: test generation, coverage gaps, flaky tests, security scans and chaos testing
A self-improving QA agent: natural-language tests for web and mobile that self-heal and remember runs
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Kane CLI (this card) | yes | 263 | |
| Playwright CLI | yes | 15,007 | drives a browser with short commands and does not write tests from requirements, while Kane generates them with coverage |
| agent-qa | no | 870 | a self-learning QA agent for web and mobile, while Kane centers on requirements, acceptance criteria and an evidence pack |
README in short
Release 0.10.0 provides macOS, Linux and Windows builds. Outputs include Markdown tests and packs under `.testmuai/evidence/`, allowing inspection of a specific failure and comparison of tested steps with requirements.