kane-cli 0.10.4

Highlights

This release makes every run observable: commands, sessions and failures now emit structured events with error codes, durations and the surface they ran on. Replay gets a self-healing read loop for screenshots and page code, and exit codes finally distinguish "your test failed" from "kane-cli broke".

[0.10.4] - 2026-11-12

Every run is now traceable

• Commands report their own lifecycle — every command emits a start and end event, including run, testmd run and explore, so a run's shape is visible even if it dies early.
• Sessions say what they are and how they ended — session start records the surface and target; session end records the outcome, including sessions that exit without finishing.
• Failures carry an error code, a domain and the actual error — no more guessing from a bare stack trace, and a testmd run that throws now logs why before closing its session.
• Ctrl+C is recorded as a signal exit, not a crash — interrupting a run no longer pollutes your history with phantom crashes.
• Telemetry is always on — there is no opt-out, and usage before login is attributed by machine id.

Exit codes you can trust

• A failed test is not a CLI error — test failures and CLI failures are now reported on separate lanes, so scripts and CI can tell them apart.
• A command that needs a login says so — instead of a generic failure, kane-cli reports the missing login explicitly.
• A run that produces no result still fails — an empty result no longer slips through as a success.

Replay that repairs itself

• Reads heal instead of giving up — screenshot and page-code reads run through the analyze v3 repair loop before failing, and a heal that finds nothing is reported as a failed read rather than a silent pass.
• Screenshot reads record the question they asked — and are checked the same way replay asks it, so verification matches intent.
• Pages are re-read after hover and drag — actions that change the page no longer leave replay working from stale content.
• query_js handles code that evaluates to a function — snippets that return a function now execute instead of erroring.
• New @manual marker — tag steps that need a human and replay treats them as replay-only, not as something to automate.

Remote and multi-test runs make sense in the logs

• testrun --remote reports dispatch, outcome and failures — and every HyperExecute VM session names the job that sent it.
• A test run coordinator is its own session — each member run names the run it belongs to and is tagged with its run type, so a parallel suite reads as one tree instead of scattered sessions.
• explore has its own lifecycle events — start, end and error are logged like any other command.

Quieter, more useful output

• Per-item success spam is gone — repeated success lines collapse into one summary, and normal states no longer emit warnings.
• Every pipeline stage reports how long it took — stages end with a duration, making slow steps easy to spot.
• Debug detail appears only when it matters — debug lines are held back and released when a run fails, and a passing test.md run drops its held step output.
• Nothing is dropped at the edges — log lines before attach, after finish and at exit are all kept.

Install

# npm
npm install -g @testmuai/kane-cli@0.10.4

# Homebrew
brew install lambdatest/kane/kane-cli

# Shell installer (latest release)
curl -fsSL https://raw.githubusercontent.com/LambdaTest/kane-cli/main/install.sh | bash

Compare: 0.10.3...0.10.4