Before: A generated suite covers only the successful login.
Negative paths are explicit work
Declared input boundaries and policy require meaningful failure and edge cases rather than a generic 'invalid input' row.
HINT for QA Engineers
Your assistant creates cases fast. HINT makes them follow the requirement.
HINT keeps the system boundary, environment, coverage rules and expected results beside test artifacts. Your assistant derives cases from what is declared instead of filling a test plan with plausible paths.
Where it helps
Three situations your team will recognize
Before: Expected results describe what the current build happens to do.
Requirements remain the oracle
Each expected outcome traces to a requirement, so a defect cannot redefine the test to make itself pass.
Before: Steps assume an environment nobody prepared.
Setup facts are declared
The SUT version, fixtures and preconditions appear before the assistant writes or updates the case.
See it
A real login test case from the demo
The Nakatomi demo derives the main vault-login case from REQ-1.
demo-nakatomi-qa/systems/vault-access/login.feature.hint ↗
# case TC-MAIN {#main_case}
Derived from REQ-1.
## precondition Approved state
Declared setup.
## step Declared actionWhat your assistant does next
It keeps the precondition and observable outcome tied to REQ-1, then the Gherkin emitter can reproduce the derivable scenario structure.
What your AI assistant receives
<case name="TC-MAIN">
Derived from REQ-1.
<precondition name="Approved state">Declared setup.</precondition>
<step name="Declared action">Perform it.</step>
</case>Your boundaries stay visible
What it will never do
- Mark a case passed without an observed result.
- Change an expected result to match faulty behavior.
- Invent an environment, fixture or requirement boundary.
FAQ
Questions qa engineers ask
Does HINT run tests?
No. It helps specify and verify the test artifacts; your existing runner executes them.
Does the emitted Gherkin work with Cucumber?
Yes. The Gherkin pack writes ordinary .feature structure; project-specific step implementations remain yours.
Can it invent expected results?
No. Expected results must come from declared requirements; otherwise the assistant reports a gap.
For your technical colleague
The vocabulary
A hintbook is a vocabulary for your profession—installed, not written by you.
plan- Sets the purpose, scope and exit conditions of a test effort.
sut- Defines the system under test and its boundaries.
case- Records a test case with a stable identifier.
precondition- States setup that must be true before a case begins.
expected- Names an observable result traced to a requirement.
testdata- Constrains fixtures and values without inventing them.
covers- Links a case to requirement or story identifiers.
defect- Structures an observed mismatch without inventing results.
Manual setup
Bootstrap is read-only: it prints instructions for the assistant. The assistant performs the installation.
npm install -g @openhint/cli
hint config
hint add @openhint/hintbook-qa-engineer
hint apply
hint verify systems/vault-access/login.feature
hint emit systems/vault-access/login.feature