From requirement to proof.
Describe what your product must do. Kane CLI designs the tests that prove it, links every test to the requirement it verifies, and measures coverage from evidence.
// The problem
A green run does not say what was covered.
Tests pass and the build ships. Nobody can say which requirements a real run has proven.
CI REPORT
- ci / lint · Successful in 18sDetails
- ci / unit-tests · 212 passed in 1m 4sDetails
- ci / build · Successful in 2m 11sDetails
- e2e / playwright · 12 passed in 3m 40sDetails
THE SAME USE CASE, BY CRITERION
- AC-1guest can add an item to the cartproven
- AC-2the total matches the cartfailed
- AC-3the receipt email is sentblocked
- AC-4a declined card shows a messageno test
CI is green, yet a run has proven only one of these four criteria.
Tests with no requirement behind them
A test checks something, but nothing records which promise in the spec it stands for.
Coverage that is guessed
A pass rate counts the tests that ran. It says nothing about criteria nobody wrote a test for.
A suite that drifts from the spec
The spec changes and the tests stay as they were. Nothing marks which ones are now out of date.
// The assurance loop
Every stage is a command you can stop and review.
No stage runs until the one before it has committed its output.
$ kane-cli context ingest
# PRDs, specs and policy pages become versioned sources.
Capture
Saves a snapshot of your requirement documents in a local store. Each snapshot is identified by a hash of its content.
// Exploration
When there is no spec, exploration builds the context.
Give the exploration agent a URL. It walks every page and flow of your app, flags what is already broken, and hands the assurance loop a starting spec. Built for legacy and agent-built apps with no documents.
- /loginmapped
- /accountsexploring
- /transfer1 finding
- /cardsmapped
- /settingsexploring
- /supportexploring
Open any browser to watch its session live.
Recorded on every page
- Controls
- Flows
- In and out links
- Screen versions
Flagged without a prompt
- UX bugTransfer stays disabled after a valid amount
- SecurityAccount number shows in the page URL
- AccessibilityCard form inputs have no labels
- A URL
- Exploration agent
- App map and findings
- Assurance loop
- Test suite
// Context
Every test traces back to a requirement document.
Kane CLI writes each test from a document your team owns, and keeps that link accurate as it changes.
// Evidence pack
Every run seals an evidence pack.
Pass or fail, one .evidence file feeds four reports: traceability, coverage, change history and root cause.
- 84 criteria
- 44 Proven
- 8 Failing
- 16 Stale
- 9 No test
- Proven
- Broken
- Not proven yet
// Your current setup
A new layer above the grid you already run.
Kane CLI checks each change in real Chrome and exports passing flows as Playwright code for your existing grid.
IN YOUR CURRENT SETUP
Kane CLInew layer
runs the change in real Chrome · seals an evidence pack
TodayNothing sits between the change and your suite, so a requirement with no test goes unnoticed.
With Kane CLIKane CLI turns each requirement into scenarios, runs them in real Chrome and seals an evidence pack.
Start with one requirement document.
Ingest a PRD, review what Kane CLI proposes, and run the tests it designs. Local runs are free.
