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
- Tweets245
- Following7
- Followers213
- Likes147
Some apps have no spec at all, and the apps that do have one end up with a spec that’s hard to maintain at the pace AI agents now ship code.
Exploration in Kane CLI starts from the app itself, so you give it a URL and Exploration walks every page, junction and flow across dozens of parallel browsers, for hours if the app needs it.
Exploration also flags UX, accessibility and security issues while it maps the app, without being told what to test.
The finished map becomes the context for Kane CLI Assurance, which turns it into use cases, scenarios, acceptance criteria and a full test suite that runs on real devices.
Build your context 🔗 testmuai.com/kane-cli/?utm_s…
#AI #KaneCLI
Kane CLI by TestMu AI retweeted
Coding agents are great. They ship code faster than any team.
The problem is the verification now. How to be sure that we are shipping what is supposed to ship and do that at machine speed.
When the same agent writes the implementation and the tests, both come from the same context window. Same assumptions, same misread of the spec. The tests assert what the code does, not what the requirement said.
That’s context bias. A model can’t flag a bug it doesn’t believe exists.
Add to that: tests run headless, one viewport, one Chromium build, mocked network. Green in CI, broken on Safari.
A passing build from a self-verified agent is a tautology, not proof.
So we built @kane_cli as an independent verification agent. Separate context, zero access to the implementation. It derives test cases from the requirement in natural language, executes them on real browsers and real devices, and logs step-level evidence: screenshots, logs, video.
Your coding agent calls it from the terminal via Kane CLI. Claude Code, Cursor, Codex, whatever you run.
Let the coding agent build. Let an independent agent verify.
Try now 👇
kane-cli 0.10.0 🚀
🧭 explore runs remote on HyperExecute, with a live map view.
🔗 Linked folders: design, review, run on a shared project.
🛡️ Policy gates judge every sealed evidence pack.
🤖 --agent and NDJSON on every command.
Explore more, prove more.
github.com/LambdaTest/kane-c…
Orders break when the payment webhook and the order page disagree. Kane CLI paid by card and checked that the webhook and the order page showed the same status, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
Shoppers who split a purchase into four payments expect a correct schedule. Kane CLI chose Pay in 4 and confirmed four interest-free payments, the first due today, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A declined pay later request can still end in a sale. Kane CLI checked that card and bank transfer were offered after the decline, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
Live from SF: the “eat with one hand, debug with the other” portion of the day. The hackathon grind is real 🔥
#KaneHackDay #KaneCLI
Floor help in action! Sometimes our agent needs a second opinion, and sometimes we do #KaneHackDay
Kane CLI by TestMu AI retweeted
Building Weave at the @kane_cli hackathon today 🧵
Asking your study group for more screen time usually means explaining yourself.
Weave gives the group a shared time budget. Going over costs everyone a little. Your reason stays private.
#KaneCLIHackDay
build with an agent, verify it with kane cli. the only npm install the crowd in the room needs today is on this standee!
#kaneclihackday #kanehackday
afternoon at kane cli hack day. everyone’s in “one more fix before 5” mode!
#kaneclihackday
kane cli hack day is live in SF 🛠️ the room is full of people building with @OpenAI, and there’s one rule: kane cli has to verify your app before you demo
submissions lock in a few hours. can’t wait to see what ships #kaneclihackday
Returning shoppers expect to pay with their saved card without typing it again. Kane CLI paid with a saved Visa, entered no card details and checked the charged amount on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A shopper paying by iDEAL leaves checkout for their bank and must return to finish. Kane CLI followed the bank redirect, approved the payment and checked the amount on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A refund is done only when the dashboard agrees. Kane CLI refunded $150 of a $250 sale and confirmed net sales fell to $100 and the status read Partially refunded, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A paid invoice that still says Unpaid sends someone chasing money that already arrived. Kane CLI pays a $480 invoice and confirms the badge changes to Paid, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
With a checkout link, the receipt is the only thing the buyer keeps. Kane CLI pays through the link and checks the payment ID, item, total and email on the receipt, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A payment link that can be paid twice ends in a refund and an awkward email to the customer. Kane CLI pays a link by card, then confirms a second payment is blocked, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
An EMI badge with the wrong monthly amount misleads a buyer who is ready to pay. Kane CLI checks where the badge shows and every monthly figure, on web, iOS Simulator or Android Emulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…
A UPI form has to reject an incomplete UPI ID before it accepts a valid one. Kane CLI tests both and follows the payment into the store's records, on web, Android Emulator or iOS Simulator.
Free to start 🔗 testmuai.com/kane-cli/?utm_s…