Repository navigation
·
1 commit
to main
since this release
Highlights
kane-cli can now run tests on cloud real devices from any machine, with a live device window in the TUI and clear messaging when a plan or platform can't provide one. Variable handling gets a proper runtime namespace — {{global.…}}, {{smart.…}} all resolve during a run — and uploaded files are reusable by alias for the whole session. Replays are steadier thanks to a stability ledger that waits for the page to settle instead of guessing.
[0.9.0] - 2026-10-15
Real-device runs
- Run on real devices from
kane-cli runor the TUI — real devices are selectable in both, and a live device window shows the session as it happens. - Real-device runs work from any machine — no local device setup required, and
--remotenow picks sensible device defaults. - You're told exactly which plan a real device needs — when the hub refuses a request on plan grounds, the message names the required plan instead of a generic failure.
- Clean refusals instead of confusing hangs — if the hub has no devices for a platform, or a saved app can't run on the chosen device, kane-cli says so and why.
- Evidence matches the device you actually got — the run record keeps the hub-allocated device name and slug, the device profile names that device, and the session's alert/permission setting is recorded on the mobile test.
Variables that resolve at runtime
{{global.…}},{{smart.…}}resolve when the objective starts — the agent gets real values, not raw placeholders.- Templates with two references work, and captures survive — multi-reference strings resolve correctly and captured values stay available for later steps.
- Smart variables now work on Windows — timezone data ships with kane-cli, so date and time variables resolve instead of erroring.
Working with uploaded files
- Reference uploads by alias anywhere in a test —
{{file.<alias>}}resolves to the session's cached copy of the file. - Aliases are yours for the whole session — a local alias is reserved up front and can't collide with another file mid-run.
Control and privacy
- Each testmd objective can set its own step cap —
max_stepsapplies per objective, with a defined fallback when it isn't set. - Live view addresses stay private — the live view address is no longer exposed, and Chrome profiles created for a run are cleaned up afterward.
Install
# npm
npm install -g @testmuai/kane-cli@0.9.0
# Homebrew
brew install lambdatest/kane/kane-cli
# Shell installer (latest release)
curl -fsSL https://raw.githubusercontent.com/LambdaTest/kane-cli/main/install.sh | bashCompare: 0.8.23...0.9.0