Kane CLI By TestMu AI

Kane CLI is a terminal-native AI testing tool from TestMu AI (formerly LambdaTest) that turns plain-English instructions into verified browser tests. Instead of writing and maintaining brittle Playwright or Selenium scripts, you simply describe what should happen — for example, "log in as an admin, open the billing page, and confirm the plan shows Enterprise" — and Kane CLI drives a real Chrome browser to execute it, verifies each step, and returns a deterministic pass or fail with a shareable evidence link. It is built for two audiences at once: humans who want fast end-to-end verification without scripting, and AI coding agents that need a reliable way to prove the code they write actually works in a browser.

Most browser automation falls into two camps: powerful but high-maintenance frameworks like Playwright and Selenium that break every time the UI shifts, or general browser agents that perform tasks without any concept of a test. Kane CLI is purpose-built for verification — every run ends in a binary pass or fail, backed by a real browser session you can replay. These are tests, not just tasks. And once a flow stabilizes, a single command exports it to native Playwright code you can drop into a maintained suite, so Kane CLI complements your existing stack instead of replacing it.

Authoring is effortless: describe flows in plain English with no selectors, no custom DSL, and no boilerplate. Everything runs in a real Chrome browser, so actions are limited to what a real user can do — there is no JS injection to bypass validation, which means results reflect reality. The same objective combined with the same context yields the same result, giving you deterministic pass or fail verdicts rather than a one-shot LLM guess.

Flakiness is handled for you. Autoheal absorbs cosmetic UI changes — renamed buttons, shifted layouts — that would break a traditional script, and pushes through to finish the journey. Vision-based dynamic waiting detects what is actually on screen, handling canvas, shadow DOM, and JS-heavy frameworks that resist selectors. Native Playwright export converts any run into editable Playwright code, and two-way migration also lets you ingest your existing Playwright or Selenium scripts.

Every run produces a shareable evidence link with full replay, video, step trace, screenshots, and network and console logs — paste the URL into Slack, Jira, or a pull request as proof. On a CAPTCHA or OTP, the built-in Ask tool pauses and asks a human to clear that one step, then resumes, bringing human-in-the-loop control without halting the workflow. Test.md captures any session as replayable markdown with variables and imports, giving you an agent-native, human-readable test artifact.

Three run modes fit every workflow: an interactive TUI for debugging, headless CLI mode for CI/CD, and agent mode with structured NDJSON output for AI agents. One command — npx @testmuai/kane-cli-skill — installs Kane CLI as a skill for Claude Code, Codex CLI, Cursor, and Gemini CLI, so your agent invokes it automatically. For pipelines, Kane CLI is CI/CD ready with standard exit codes (0 pass, 1 fail, 2 error, 3 timeout) and support for GitHub Actions, GitLab CI, Jenkins, and Bitbucket Pipelines. By default it runs locally on Chrome, or it can connect to TestMu AI's remote grid for cross-browser execution at scale.

Kane CLI is made for developers and vibe coders verifying features end-to-end before a pull request without writing 200 lines of test code, for QA engineers who want resilient checks that do not shatter on every UI tweak, for AI coding agents that need to close the loop between writing code and proving it works, and for DevOps and platform teams running browser checks in CI without a test codebase to maintain.