scenarios-generatorlisted
Install: claude install-skill YevheniiaQA/qa-agent-toolkit
# Scenarios Generator
This is the stage where test quality is actually decided. Automation code
generated later can only be as good as the scenarios it implements, and a
scenario nobody can trace to a requirement is a scenario nobody can defend in a
release meeting.
Scenario JSON is the single source of truth for the whole pipeline. TMS cases and
automated tests are both projections of it, which is how they stay in sync.
## Test level selection — follow the pyramid
| Requirement lives at | Generate | Why |
|---|---|---|
| User Story | End-to-end scenarios | Covers the user's actual path across components |
| Backend Task | API scenarios | Fast, stable, exhaustive on input classes |
| Frontend Task | UI component scenarios | Rendering, states, validation, accessibility |
| Cross-service | Integration scenarios | Contract boundaries and failure modes |
Push volume down the pyramid. Exhaustive input-class coverage belongs at the API
level; end-to-end scenarios cover representative happy paths and the few failure
paths that only manifest across components. A suite with forty e2e scenarios and
four API scenarios is a suite that will be flaky and slow.
## Apply ISTQB techniques explicitly
Every scenario names the technique that produced it — equivalence partitioning, boundary value
analysis, decision table, state transition, informed error guessing — so a reviewer can tell
whether coverage is adequate rather than merely plentiful. Read `references/techniques.md`
for when to