create-test-cases

Create Test Cases Skill

Generates mobile UI test cases for automated visual and interaction testing, scoped to a specific Jira task, PRD, Figma design, or feature description.

Scope Restriction

ONLY generate test cases for mobile app UI.

• ✅ Screen states (loading, empty, error, success)
• ✅ User interactions (tap, swipe, scroll, type, long-press)
• ✅ Navigation flows between screens
• ✅ Visual element assertions (visible, hidden, text content, enabled/disabled)
• ✅ Platform-specific behavior (Android vs iOS)
• ❌ API endpoint testing
• ❌ Web/browser testing
• ❌ Backend/service testing
• ❌ Database testing
• ❌ Performance/load testing

Input Sources (Priority Order)

1. Jira Ticket URL

When the user provides a Jira ticket URL:

1. Fetch the ticket using mcp_atlassian_getJiraIssue.
2. Extract: summary, description, acceptance criteria, subtasks, linked issues.
3. If the ticket references a Confluence PRD/RFC, fetch it using mcp_atlassian_getConfluencePage.
4. Generate test cases scoped to the ticket's requirements.

2. PRD / Confluence Document

When the user provides a Confluence page URL or pastes PRD content:

1. Fetch the page using mcp_atlassian_getConfluencePage if URL.
2. Parse requirements, user stories, and acceptance criteria.
3. Generate test cases covering every stated requirement.

3. Figma Design

When the user provides a Figma file/frame URL:

1. Fetch design screenshot using mcp_figma_dev_mod_get_screenshot.
2. Extract component structure using mcp_figma_dev_mod_get_design_context.
3. Identify UI states, variants, and interaction flows from the design.
4. Generate test cases for:
   • Each visible screen state (default, loading, empty, error, success)
   • Component interactions (buttons, inputs, toggles, selections)
   • Navigation flows between screens
   • Visual correctness assertions (element presence, positioning, content)
5. Cross-reference with Jira/PRD if available for requirement traceability.

4. Free-Text Description

When the user describes the feature or task directly:

1. Parse the description for requirements and constraints.
2. Ask clarifying questions only if critical information is missing.
3. Generate test cases from the stated scope.

Workflow

Input (Jira / PRD / Figma / Text)
        │
        ▼
Fetch & Parse Context
        │
        ▼
Identify Test Scope
  - Features & requirements
  - UI states from Figma
  - User roles & permissions
  - Input boundaries & edge cases
  - Error & failure scenarios
        │
        ▼
Generate Test Cases
  - Happy paths
  - Negative / error paths
  - Edge cases & boundaries
  - Visual state assertions
  - Platform-specific (if any)
        │
        ▼
Output: Jira Comment + CSV File

Test Case Structure

Each test case must include:

Field	Description
ID	Sequential (TC001, TC002, ...)
Title	User able to ... (happy) / User not able to ... (negative)
Preconditions	Device, OS version, user role, screen context, navigation state
Steps	Numbered mobile UI actions (tap, swipe, scroll, type, navigate)
Expected Result	Observable visual outcome (element visible/hidden, text, screen state)
Category	Smoke or Regression
Priority	P0 (Critical), P1 (Important), P2 (Nice-to-have)
Tags	UI, Visual, Navigation, Interaction, Negative, Boundary, Permission, Offline, Platform

Category Assignment Rules

Smoke — Assign when ALL conditions are met:

• Core happy path flow (the primary user journey)
• Feature is fundamentally broken if this test fails
• No edge cases, boundary conditions, or error scenarios
• Minimum viable interaction to validate the feature works
• Typically P0 priority

Regression — Assign for everything else:

• Edge cases and boundary tests
• Negative scenarios and error handling
• Permission and offline failures
• Platform-specific edge behaviors
• Secondary or alternative flows

Output Format

Jira Comment

Post via mcp_atlassian_addCommentToJiraIssue with contentFormat: markdown:

## Test Cases: {Ticket Title}

*Auto-generated by qa-test-case-generator | Source: {source}*

### Feature Area: [Name]

#### Smoke Tests (Happy Path)

| ID | Title | Preconditions | Steps | Expected Result | Category | Priority | Tags |
|----|-------|---------------|-------|-----------------|----------|----------|------|
| TC001 | User able to ... | - Precondition | 1. Step 1\n2. Step 2 | Expected behavior | Smoke | P0 | UI |

#### Regression Tests

| ID | Title | Preconditions | Steps | Expected Result | Category | Priority | Tags |
|----|-------|---------------|-------|-----------------|----------|----------|------|
| TC003 | User not able to ... | - Precondition | 1. Step 1\n2. Step 2 | Error state visible | Regression | P1 | Negative |

CSV Files

Directory: /test-cases/

Full CSV: {epic_key}_{short_desc}_test_cases.csv Smoke CSV: {epic_key}_{short_desc}_smoke_tests.csv

Columns (both files):

Ticket,Ticket Title,Feature Area,TC ID,Title,Preconditions,Steps,Expected Result,Category,Priority,Tags

Rules:

• One row per test case
• Escape commas and quotes within fields
• Include header row
• Sort by Ticket, then TC ID
• Full CSV contains all test cases (Smoke + Regression)
• Smoke CSV contains only Category=Smoke test cases

IMPORTANT — Direct file writing only: Write CSV content directly using create_file. Do NOT create Python scripts, shell scripts, or any helper programs to generate the output. The agent has all the data needed and must write the files itself in a single step.

Coverage Directives

Requirement Coverage

• Every requirement and acceptance criterion must have at least one test case.
• Field validations get individual test cases per rule.

Edge Cases & Negative Scenarios

For every happy path, generate:

• Invalid inputs: empty fields, special characters, max length in text inputs
• Boundary conditions: min/max values in form fields, empty lists, single item
• Permission failures: unauthorized screen access, disabled UI elements
• Network failures: offline state, loading timeout (visual indicators)
• State transitions: back navigation, app backgrounding, screen rotation
• Platform differences: Android back button vs iOS swipe-back

Mobile UI Visual Focus

Prioritize assertions on:

• Screen navigation correctness (correct screen rendered after action)
• Widget visibility after state changes (loading → content → error)
• Loading states (skeleton/shimmer visible, replaced by content on success)
• Empty states (illustration + message visible when no data)
• Error states (error widget visible + retry button + toast messages)
• Form interactions (field focus, inline validation messages, submit button state)
• Pull-to-refresh behavior (indicator visible, content refreshes)
• Scroll & pagination (load-more indicator, new items appended)
• Platform differences (Android Material vs iOS Cupertino)
• Deep linking (correct screen opens from external link)
• Bottom sheet / modal behavior (appears, dismissible, content correct)
• Tab navigation (active state, content switches)
• Gesture interactions (swipe-to-delete, drag-to-reorder)

Figma-Driven Assertions

When Figma designs are available:

• Assert component presence matching mobile design specs
• Verify text content matches design copy on mobile screens
• Validate state variants (pressed, disabled, selected, focused)
• Verify icon and illustration presence per screen
• Validate spacing and layout per mobile breakpoint
• Check component variants across light/dark mode
• Verify bottom navigation and tab bar states

Priority Guidelines

Priority	Criteria
P0	Core functionality, data integrity, security, blocking flows
P1	Important UX, secondary features, non-blocking validation
P2	Cosmetic, extreme edge cases, nice-to-have behaviors

Post-Generation Checklist

• Every acceptance criterion has at least one TC
• No duplicate or overlapping test cases
• All preconditions are concrete
• Steps are numbered and unambiguous
• Expected results are observable
• Negative cases exist for every happy path
• Priority distribution is realistic
• Tags are correctly assigned
• Figma states covered (if design provided)
• Smoke tests cover only core happy paths (no edge/boundary/negative)
• Every feature area has at least one Smoke test case
• Smoke CSV generated alongside full CSV