home / skills / wellapp-ai / well / test-hardening

This skill converts verified QA criteria into automated tests, turning passing scenarios into regression coverage to safeguard futures.

npx playbooks add skill wellapp-ai/well --skill test-hardening

Review the files below or copy the command above to add this skill to your agents.

Files (1)
SKILL.md
5.2 KB
---
name: test-hardening
description: Convert passed QA Contract criteria to automated tests
---

# Test Hardening Skill

Convert verified QA Contract criteria (G#N, AC#N) into permanent automated tests. Ensures passing scenarios become regression tests.

## When to Use

- After qa-commit returns GREEN for a commit
- After debug skill fixes an issue (Phase 7: Harden)
- Before pushing PR (ensure all criteria have tests)
- Manually with "use test-hardening skill"

## Input: Verification Report

From qa-commit's Verification Report:
- List of passed G#N (Gherkin scenarios)
- List of passed AC#N (acceptance criteria)

## Phase 1: Analyze Criteria

### 1.1 Categorize by Test Type

| Criteria Type | Test Framework | Location |
|---------------|----------------|----------|
| G#N (Backend) | Jest | `apps/api/**/*.test.ts` |
| AC#N (UI State) | Storybook | `**/*.stories.tsx` |
| AC#N (Interaction) | Playwright | `tests/e2e/**/*.spec.ts` |

### 1.2 Check Existing Tests

```
Grep: "G#[N]" or "[scenario name]" in test files
```

Skip if test already exists.

## Phase 2: Generate Backend Tests (G#N)

For each passed G#N without existing test:

### 2.1 Template

```typescript
// apps/api/src/[feature]/__tests__/[feature].test.ts

describe('[Feature Name]', () => {
  // G#1: [Scenario name]
  it('should [expected behavior]', async () => {
    // Arrange
    const input = { /* test data */ };
    
    // Act
    const response = await request(app)
      .[method]('[endpoint]')
      .send(input);
    
    // Assert
    expect(response.status).toBe([status]);
    expect(response.body).toMatchObject({ /* expected */ });
  });

  // G#2: [Scenario name]
  it('should return [error] when [condition]', async () => {
    // Test implementation
  });
});
```

### 2.2 Generate Test

1. Extract endpoint, method, expected response from G#N
2. Create test file if not exists
3. Add test case with G#N reference in comment
4. Run test to verify it passes

```bash
npm run test -- --grep "[scenario name]"
```

## Phase 3: Generate Storybook Stories (AC#N - States)

For state-based AC#N:

### 3.1 Template

```typescript
// apps/web/src/[component]/[Component].stories.tsx

import type { Meta, StoryObj } from '@storybook/react';
import { Component } from './Component';

const meta: Meta<typeof Component> = {
  title: 'Features/[Feature]/[Component]',
  component: Component,
};

export default meta;
type Story = StoryObj<typeof Component>;

// AC#1: Renders without error
export const Default: Story = {
  args: { /* default props */ },
};

// AC#2: Shows loading state
export const Loading: Story = {
  args: { isLoading: true },
};

// AC#3: Shows error state
export const Error: Story = {
  args: { error: 'Something went wrong' },
};

// AC#4: Shows empty state
export const Empty: Story = {
  args: { data: [] },
};
```

### 3.2 Generate Story

1. Check if stories file exists
2. Add missing story variants for each AC#N
3. Run Storybook to verify renders

```bash
npm run storybook -- --smoke-test
```

## Phase 4: Generate E2E Tests (AC#N - Interactions)

For interaction-based AC#N:

### 4.1 Template

```typescript
// tests/e2e/[feature].spec.ts

import { test, expect } from '@playwright/test';

test.describe('[Feature Name]', () => {
  // AC#5: User can submit form
  test('should allow form submission', async ({ page }) => {
    await page.goto('/[route]');
    
    await page.fill('[name="field"]', 'value');
    await page.click('[type="submit"]');
    
    await expect(page.locator('.success')).toBeVisible();
  });

  // AC#6: Keyboard navigation works
  test('should support keyboard navigation', async ({ page }) => {
    await page.goto('/[route]');
    
    await page.keyboard.press('Tab');
    await expect(page.locator(':focus')).toHaveAttribute('name', 'first-field');
  });
});
```

### 4.2 Generate Test

1. Check if E2E test file exists
2. Add test case for each interaction AC#N
3. Run Playwright to verify

```bash
npx playwright test [feature].spec.ts
```

## Phase 5: Verify Tests Pass

Run all generated tests:

```bash
# Backend
npm run test

# Storybook
npm run storybook -- --smoke-test

# E2E (if applicable)
npx playwright test
```

## Phase 6: Update Test Summary

```markdown
## Test Hardening Report

### Generated Tests

| Criteria | Type | File | Status |
|----------|------|------|--------|
| G#1 | Jest | `[path]` | CREATED/EXISTS |
| G#2 | Jest | `[path]` | CREATED/EXISTS |
| AC#1 | Storybook | `[path]` | CREATED/EXISTS |
| AC#3 | Playwright | `[path]` | CREATED/EXISTS |

### Test Results

| Suite | Total | Passed | Failed |
|-------|-------|--------|--------|
| Jest | [N] | [N] | 0 |
| Storybook | [N] | [N] | 0 |
| Playwright | [N] | [N] | 0 |

### Coverage Update

- Backend: [N]% → [N]%
- Frontend: [N]% → [N]%
```

## Integration with Debug

When invoked from debug skill Phase 7 (Harden):

1. Receive the reproduction steps from debug
2. Create regression test to prevent recurrence
3. Add test with reference to original issue

```typescript
// Regression test for [issue description]
// Debug session: [date]
it('should not [bug behavior] when [condition]', async () => {
  // Reproduction steps from debug
});
```

## Invocation

Invoked by:
- `qa-commit` - After GREEN verdict
- `debug` - Phase 7 Harden
- Push-pr mode - Pre-push verification

Or manually with "use test-hardening skill".

Overview

This skill converts verified QA Contract criteria (G#N and AC#N) into permanent automated tests so passing scenarios become regression checks. It produces Jest backend tests, Storybook state stories, and Playwright end-to-end specs, then runs the suites to confirm tests pass.

How this skill works

The skill ingests a Verification Report listing passed G#N scenarios and AC#N acceptance criteria. It categorizes each criterion by test type, checks for existing tests, and generates missing Jest tests, Storybook stories, or Playwright specs using standardized templates. Finally it runs the relevant test runner commands and produces a Test Hardening Report summarizing created files and results.

When to use it

  • After qa-commit returns GREEN for a commit
  • After a debug session fixes an issue (Phase 7: Harden)
  • Before merging a PR to ensure all verified criteria have tests
  • Manually when you want to convert verified criteria into regressions
  • During release prep to lock verified behaviors into tests

Best practices

  • Include the original criterion ID (G#N or AC#N) as a comment in each generated test
  • Prefer small, focused tests that assert one behavior per case
  • Skip generation when an existing test already references the scenario or criterion
  • Run the specific test runner (Jest/Storybook/Playwright) after generation to ensure tests are passing
  • Add a brief regression note linking to the debug session or issue when hardening a bug fix

Example use cases

  • A backend API G#3 that returned correct invoice totals is converted into a Jest test under apps/api/**/__tests__
  • A UI AC#2 that describes the loading state is added as a Storybook story variant for the component
  • An interaction AC#5 (form submission) becomes a Playwright spec in tests/e2e to prevent regressions
  • After a debug session reproduces a flaky invoice-parsing bug, generate a regression test referencing the debug steps
  • Before pushing a release, run the skill to ensure every green QA criterion has an automated test

FAQ

What inputs does the skill require?

A Verification Report listing passed G#N scenarios and AC#N acceptance criteria, plus access to the repository to add files.

Which frameworks and file locations are used?

Backend G#N → Jest (apps/api/**/*.test.ts). State AC#N → Storybook stories (**/*.stories.tsx). Interaction AC#N → Playwright (tests/e2e/**/*.spec.ts).

How does it avoid duplicating tests?

It searches test files for the scenario or criterion marker and skips generation if a matching test or story already exists.