Generate structured, non-redundant manual test cases ready for Xray import, with optional JDD test data.
Workflow
Step 1 — Analyze Input
Read the user story, feature description, or acceptance criteria. Identify:
- Testable requirements and acceptance criteria
- Input fields, workflows, business rules
- Risk areas (data integrity, security, UX)
- Ambiguities to flag
Step 2 — Select Techniques
Apply techniques based on what the input contains. See test design techniques.
| Input characteristic | Technique |
|---|---|
| Any input field | Equivalence Partitioning |
| Numeric/date/length fields | Boundary Value Analysis |
| Multiple conditions combined | Decision Tables |
| Multi-step workflow or statuses | State Transition |
| 3+ independent parameters | Pairwise Testing |
| Past defects, common failures | Error Guessing |
Step 3 — Design Test Cases
For each requirement, generate test cases covering: nominal, boundary, invalid input, error handling, and business rules. Deduplicate aggressively — quality over quantity.
Step 4 — Generate CSV
Output a semicolon-separated CSV table. See CSV format guide and examples.
Columns (9): Test ID;Summary;Description;Repository;Preconditions;Test Steps;Expected Results;Priority;Labels
Step 5 — Generate JDD (optional)
When test data is complex or the user requests it, generate a JDD JSON file. See JDD generation guide.
Defaults (overridable)
| Setting | Default |
|---|---|
| Language | French |
| CSV separator | ; (semicolon) |
| Test ID format | TC-[MODULE]-[NNN] |
| Test name format | TC_[PRODUCT]_[MODULE]_[FEATURE]_[ID]_[NAME] |
| Priority levels | P0 = blocker/smoke, P1 = functional/regression, P2 = comfort/edge |
| Repository path | [Module]/[SousModule]/[Fonctionnalité]/ |
Priority Definitions
- P0 — Blocking path. If this fails, the feature is unusable. Smoke tests, critical business rules.
- P1 — Core functional. Regression-worthy scenarios, main error handling, key validations.
- P2 — Edge cases, cosmetic validations, comfort features, boundary exploration.
Constraints
- ALL test content in French
- No duplicate test cases — merge overlapping scenarios
- No vague steps — every step has a concrete action and observable expected result
- Use realistic test data (domain-appropriate values, not "test123")
- Each test traces to a requirement or acceptance criterion
Integration
The CSV output can feed into the qa-automation skill to scaffold automated Playwright tests.