End-to-end flows on your local browser, powered by natural language. Instant validation. Deep bug discovery. Automation that ships with confidence.

Joined May 2026
12 Photos and videos
kane-cli 0.4.4 ๐Ÿš€ โœจ One place to set your start URL ๐ŸŒŸ Clearer errors when a URL is missing ๐Ÿ’ซ Fewer surprises in long sessions ๐ŸŽ‰ Linux arm is now supported Sharper than last time. github.com/LambdaTest/kane-cโ€ฆ
3
3
22
kane-cli 0.4.3 ๐Ÿ›ก๏ธ ๐Ÿ” New browser-automation tools ๐ŸŒŸ Isolated clipboard for the agent ๐Ÿ” Set Auth header via Set cookies โœจ Contextual hints in the footer Tests that survive. github.com/LambdaTest/kane-cโ€ฆ
1
4
141
Kane CLI by TestMu AI retweeted

2
21
29
342
One objective. Three ways to run it. Same natural language objective, all three Kane CLI modes: ๐Ÿ–ฅ๏ธ TUI Mode: type it, watch the step tree fill in. State persists, so the next run picks up where you left off. โšก CLI: kane-cli run "..." executes end to end, records the session, saves to KaneAI. ๐Ÿค– Agent Mode: Claude Code calls --agent --headless and reads structured step events directly. Same binary. Same syntax. Your desk, a pipeline, or an agent's loop. The agent builds it. Kane CLI proves it works. Get started ๐Ÿ”— testmuai.com/kane-cli/?utm_sโ€ฆ #BrowserAutomation #KaneCLI
6
9
110
Half your CI minutes go to fixing tests that broke because someone changed a color. ๐ŸŽจ That is not a test suite. That is a tax.๐Ÿงฎ
16
17
80
$ ๐™บ๐šŠ๐š—๐šŽ ๐™ฒ๐™ป๐™ธ --๐š๐š’๐š™ ๐Ÿถ4๐Ÿ’ก Add --agent, and the output becomes NDJSON, one event per line. Grab just the verdict: ๐š”๐šŠ๐š—๐šŽ-๐šŒ๐š•๐š’ ๐š›๐šž๐š— "..." --๐šŠ๐š๐šŽ๐š—๐š 2>/๐š๐šŽ๐šŸ/๐š—๐šž๐š•๐š• | ๐š๐šŠ๐š’๐š• -1 | ๐š“๐šš -๐š› .๐šœ๐š๐šŠ๐š๐šž๐šœ Benefit: your AI agent reads pass/fail & every value you stored, no log scraping
1
4
24
86
New in 0.4.2: attach files to the AI test case generator. Two ways to do it. ๐Ÿ–ฅ๏ธIn the TUI: @-mention a file in your prompt. Kane reads it and writes the scenarios: signup flows, validation, XSS checks, password security. /save and they're runnable. โšกIn one shot: kane-cli generate --files "your-file" with your objective. No session, no clicks. Built for scripts and pipelines. Your file is the spec. Try now๐Ÿ”—testmuai.com/kane-cli/?utm_sโ€ฆ #KaneCLI
1
9
11
134
$ ๐™บ๐šŠ๐š—๐šŽ ๐™ฒ๐™ป๐™ธ --๐š๐š’๐š™ ๐Ÿถ๐Ÿน๐Ÿ’ก Building with Claude Fable 5: Every regeneration shifts the UI. Buttons move, fields get renamed, and selector-based tests break on every pass. Write your checks as intent instead. "Sign up, then confirm the dashboard loads" It survives the churn because it looks for the outcome, not a hard-coded selector. #KaneCLI
19
23
168
Claude Fable 5 built a Replit clone from a prompt in minutes.๐Ÿคฏ But what about verifying the build? When it comes to actually testing such a massive build and matching the pace of AI generation, only resilience can keep up. And guess who has that resilience?๐Ÿ˜Ž
19
24
40
338
kane-cli 0.4.2 ๐Ÿ›ก๏ธ โœจ Attach files to generate sessions ๐ŸŒŸ Smarter generate-mode interaction ๐Ÿ“ฆ Install subcommand ๐Ÿ’ซ Project and folder gate in generate mode Faster setup, fewer surprises. github.com/LambdaTest/kane-cโ€ฆ
8
10
183
Every team eventually hits the same wall: code velocity outruns verification. ๐Ÿงฑ Ask any dev you know. Shipping the feature is the fast part now. Proving it actually works in a browser is what quietly eats the afternoon. And when nobody has time for that, the proof just never happens. So we handed that part to Kane CLI. It lives inside our GitHub Actions. Every push, every PR, opens a real Chrome browser and walks the flows that actually matter: login, checkout, signup, the ones that hurt when they break. Written in plain English, not selectors. If something is off, we hear it before merge, not after release. Latest run ๐Ÿ‘‡ โœ… 4 passed, 0 failed ๐Ÿ“ธ screenshots and logs on every step We stopped shipping on "looks fine" and started shipping with proof. Test every push in a real browser. ๐Ÿ‘‰ bit.ly/4xkCJwG #CICD #DevOps #KaneCLI
23
24
189
$ ๐™บ๐šŠ๐š—๐šŽ ๐™ฒ๐™ป๐™ธ -- ๐š๐š’๐š™ ๐Ÿถ๐Ÿธ๐Ÿ’ก The TUI keeps one browser session alive across your runs. That's persistence. Run 1: "go to wikipedia, click English" Run 2: "store the title, assert it contains 'Wikipedia'" No navigation. It's still on the page. The page, cookies, and your logins all carry over. #KaneCLI
10
10
190
The state of browser automation in 2026 ๐Ÿ‘‡ ๐Ÿข Selenium: 200 lines selectors that die on every redesign ๐Ÿ”ง Playwright: setup, fixtures, endless maintenance โšก Kane CLI: "go to the site and check if checkout works" Pick your fighter๐ŸฅŠ #BrowserAutomation #Developers #QA #TestMuAI #KaneCLI
2
8
10
264
One prompt in Lovable. A full waitlist landing page, live in minutes. The question that decides everything: does it actually work for a real user? That's the job Kane CLI does. In this run, it: โ†’ Opened the published app in a real Chrome browser. โ†’ Typed an email and clicked Join Waitlist. โ†’ Confirmed the success message appeared. โ†’ Returned pass, with a video and a step trace. This is what agent-native verification looks like. Your coding agent ships the feature. Kane CLI confirms it works in the browser, the way a user would see it. Code is fast. Verification used to be slow. Not anymore. The agent builds it. โฌ‡๏ธInstall Now ๐š—๐š™๐š– ๐š’๐š—๐šœ๐š๐šŠ๐š•๐š• -๐š @๐š๐šŽ๐šœ๐š๐š–๐šž๐šŠ๐š’/๐š”๐šŠ๐š—๐šŽ-๐šŒ๐š•๐š’ OR ๐š‹๐š›๐šŽ๐š  ๐š’๐š—๐šœ๐š๐šŠ๐š•๐š• ๐™ป๐šŠ๐š–๐š‹๐š๐šŠ๐šƒ๐šŽ๐šœ๐š/๐š”๐šŠ๐š—๐šŽ/๐š”๐šŠ๐š—๐šŽ-๐šŒ๐š•๐š’ #VibeCoding #AI #BrowserTesting #DevTools #Lovable #KaneCLI #TestMuAI
1
6
8
244
Kane CLI by TestMu AI retweeted
๐Ÿค– I just watched an AI agent (Kane CLI) test my checkout flow without writing a single selector or line of code. I described the test in natural language: go to the website, click Buy Pro, and confirm the total shows $19.00. The agent opened a real browser, ran the flow, and reported pass or fail right in my terminal. Give it a try for your workflow: ๐Ÿš€๐Ÿ‘‡ testmuai.com/kane-cli/ #KaneAI #KaneCLI #AITestAuthoring #NLPTesting
7
9
165
Kane CLI by TestMu AI retweeted

26
32
548
Kane CLI by TestMu AI retweeted
Browser agents stop at a screenshot and assume it worked. Kane CLI searched GitHub for React, opened the repo, read 246k stars, then proved it's the real official repo, not a fork. Agents do the task. Kane proves it. โญ Star : github.com/LambdaTest/kane-cโ€ฆ #KaneCLI #DevTools #QA
3
6
192
New daily series โ†’ $ ๐™บ๐šŠ๐š—๐šŽ ๐™ฒ๐™ป๐™ธ --๐š๐š’๐š™๐Ÿ’ก Every day, one tip on getting more out of Kane CLI. Starting with the basics, working up to the good stuff. First one's in the thread. ๐Ÿ‘‡ #KaneCLI #TestMuAI
1
1
1
56
$ ๐™บ๐šŠ๐š—๐šŽ ๐™ฒ๐™ป๐™ธ --๐š๐š’๐š™ ๐Ÿถ๐Ÿท๐Ÿ’ก Kane CLI objectives = 3 building blocks: > Do something: "click the Login button" > Grab something: "store the price as price" > Check something: "assert the page shows Welcome" Chain them in one English line. Keep it under ~15 steps. That's the whole api. #KaneCLI
6
8
95