Design detailed reproducible test cases (Given/When/Then, one check per case); not smoke/regression checklists
Install
npx skillscat add ozerohax/assistagents/testing-test-case Install via the SkillsCat registry.
SKILL.md
Link to requirement/task/PRD
Acceptance criteria / expected behavior
Key user scenarios list
Requirement/AC identifiers (if any)
Scenario priority/criticality
Known bugs/risks list
Environment and data constraints
</input_requirements>
This skill describes how to design test cases; documenting them is optional
If documentation is still needed, the format can be used as a template
Use Given-When-Then
Separate positive/negative scenarios
One case = one check
Apply equivalence partitioning and boundary values
State preconditions and postconditions
Separate critical branches into their own cases
Separate test data from steps
</test_design>
Steps are reproducible without interpretation
Expected result is verifiable and unambiguous
No duplicates or overlaps
Edge cases and validation errors are covered
Test data and initial state are stated
There is a link to the requirement/AC or acceptance criterion
Documenting cases is optional, but the check logic must be explicit
</quality_rules>
Do not mix multiple scenarios in one case
Do not use vague wording
Do not include more than one requirement/AC in one case
Do not leave implicit assumptions about data
</do_not>
TC-LOGIN-001: Valid login with existing user
Clear scenario, expected outcome is verifiable
Check login works
No steps and no expected result