Loading
Preparing the page
Public context is loading before any interactive tool code.
Preparing the page
Public context is loading before any interactive tool code.
Structure explicit happy, negative and boundary scenarios into numbered Markdown test cases without inventing product behavior.
Your checked result will appear here.
The happy, negative and boundary lanes remain distinct while sharing the same preconditions and trace. Missing coverage cannot hide inside a polished paragraph.
Every scenario and expected result must be supplied. MCXAI adds numbering, shared preconditions, case labels and the requirement trace.
Use a testable outcome with enough detail to decide whether an observed result passes.
Record the state that must exist before each case begins, without embedding credentials or customer data.
At least one happy, negative and boundary scenario is required before generation can continue.
Copy the result or download Markdown, then attach automation and execution evidence in the real test system.
This tested brief supplies every expected result. The generated matrix labels cases TC-001 through TC-003 and repeats the exact rejection requirement under each trace.
{
"feature": "CSV import",
"requirement": "Rows without an email address are rejected with a field-level message.",
"preconditions": [
"The reviewer can open the import screen"
],
"cases": [
{ "kind": "happy", "scenario": "Upload one row with a valid email", "expected": "The row is accepted" },
{ "kind": "negative", "scenario": "Upload one row with no email value", "expected": "The row is rejected and email is identified" },
{ "kind": "boundary", "scenario": "Upload one row with a whitespace-only email", "expected": "The value is treated as empty and rejected" }
]
}TC-001 — Happy path TC-002 — Negative path TC-003 — Boundary
Observed: 3 cases · 1 happy · 1 negative · 1 boundary
happy · negative · negativeThe matrix stops until at least one explicit boundary scenario is supplied.
A category label does not guarantee useful coverage. Add cases that reflect real data, permissions, failure recovery and system boundaries.
Cover retries, partial completion, duplicate submissions and interruption between steps.
Verify allowed and disallowed roles separately; never infer access rules from a happy path.
Include empty, missing, duplicated, oversized, multilingual and malformed values where relevant.
Specify the user message, log, metric or audit record required when a case fails.
The output documents intended checks. It does not run code, inspect coverage or certify that the implementation passed.