Repository navigation
kane-cli 0.6.8
Highlights
A new options-first interaction model replaces the old composer and slash palette for assurance, making test design faster and more keyboard-driven. The gaps view becomes a dual-axis nested tree — separating what's designed from what's proven — with risk-weighted scoring throughout. Chained design runs now narrate, ask, and re-steer reliably, with stable sequential IDs and rich dossier views surfaced across the graph.
[0.6.8] - 2026-07-28
A new options-first design surface for assurance
- Composer and slash palette are gone — the new options-first surface replaces them with a shared panel family; answers, recommendations, and multi-select actions are all reachable from one place without typing slash commands.
- Question panel shows your picks and lets you act on them in bulk — visible selections, real multi-select, and a
» Go with recommendationsbatch action (reachable via Tab) make answering faster on long option lists. - Design check-in is phase-aware — the headline updates per phase, labels switch between "Design view" and "✎ Edit", and a "Continue to next" prompt appears at the right moment instead of leaving you to guess what comes next.
- Typed answer API with go-with-recommendations — answers are now strongly typed end-to-end; accepting the recommended set is a single action rather than manual selection.
Chained design runs that actually work
- A chained design can ask questions and be steered mid-run — the nested run channel carries the conversation forward; the child run no longer parks silently on IDLE.
- Re-design is now an explicit act — a chained child narrates its progress and asks as expected; re-entering design mode requires a deliberate step, not an accidental state.
- Extract-design chaining has a full picker and ask-again loop — the picker, bridge, and ask-again flow are wired together so you can keep refining without restarting.
- Bar returns to the session after a chained run — the status bar correctly resumes on the parent session once the child finishes; the ask-again path can also open the view.
Gaps becomes a dual-axis tree
- Gaps now shows designed vs. proven as separate axes — the nested tree splits coverage into what's been designed and what's been execution-proven, so gaps are visible on both dimensions at once.
--jsonand--fromflags now reach the subcommand — a parent-level flag-stealing bug is fixed; both flags work correctly when passed togaps.- Risk-first ordering throughout — tables and item lists sort by risk weight, so the most consequential gaps surface first.
Stable IDs and richer test identification
- Every node gets a sequential ID, visible everywhere — sequential per-kind logical IDs are minted at all four creation points, so rows, items, and test files all carry consistent, human-readable identifiers instead of fabricated labels.
- Failing rows name the sealed test by ID and basename — when a row fails, the display shows the real test ID and file stem; names are never fabricated.
- Descriptive test-file stems — generated test files are named from the test's logical identity, making them easier to find in your file system.
- Sticky display names with reserved namespace and case tolerance — display names persist correctly across state transitions and handle mixed-case input without collision.
Dossier and graph views
- Full dossier view across the graph seam — an all-info dossier surfaces every relevant fact for a node, including chain-union evidence and live-successor status.
- Scoped explorer opens on the state's minted items — when you open the explorer from a given state, it starts on that state's items rather than the full list.
- Dossier actually loads under ESM — a
requirecall under an ESM module was silently breaking the dossier; fixed.
Accuracy and display fixes
- Unchecked checkboxes no longer render as
[checked]— CDP tri-state normalization now correctly distinguishes unchecked from checked. - Receipts stay in the chat gutter — wrapped receipt lines no longer snap to column 0; they hang correctly in the gutter margin.
- Agent prose-stop mid-design is not treated as completion — if the agent stops writing mid-design, the reply-first panel appears instead of the run being silently marked done.
- OR checkpoints wait for every required store key — previously only the first key in an OR checkpoint was awaited; all keys are now required before the checkpoint clears.
- Formula pipeline is consistent — window cuts apply before failure attribution; nameless nodes never render a cell ID; the latest-only cut is honored in strict-mode failure checks.
Install
# npm
npm install -g @testmuai/kane-cli@0.6.8
# Homebrew
brew install lambdatest/kane/kane-cli
# Shell installer (latest release)
curl -fsSL https://raw.githubusercontent.com/LambdaTest/kane-cli/main/install.sh | bashCompare: 0.6.7...0.6.8