QA Plan Generation

Generate a structured test plan from an existing PRD. Every test scenario traces back to a functional requirement and acceptance criterion.

Process

1. Read PRD

• Look for PRD in docs/prd/ in the current project
• If not found, ask the user for the PRD location
• Extract: Epics, FRs, NFRs, acceptance criteria

2. Generate Test Scenarios

For each acceptance criterion, generate test scenarios across these categories:

3. Classify Each Scenario

Mark each test scenario as:

• Manual — requires human judgment (UX, visual, exploratory)
• Automated — deterministic, repeatable (unit, integration, API)
• Deferred — not testable yet (infrastructure not ready, out of current scope)

4. Save Test Plan

Save to docs/qa/<project-name>-test-plan.md in the project repo.

Output Format

# Test Plan: [Project Name]

> **PRD:** docs/prd/<project-name>-prd.md
> **Date:** YYYY-MM-DD
> **Coverage:** X scenarios across Y epics

## EPIC-001: [Title]

### FR-001: [Requirement]

| # | Scenario | Category | Type | Pass Criteria | Status |
|---|----------|----------|------|---------------|--------|
| TC-001 | [Description] | Happy Path | Automated | [Expected result] | [ ] |
| TC-002 | [Description] | Edge Case | Automated | [Expected result] | [ ] |
| TC-003 | [Description] | Error | Manual | [Expected result] | [ ] |

### FR-002: [Requirement]

| # | Scenario | Category | Type | Pass Criteria | Status |
|---|----------|----------|------|---------------|--------|
| TC-004 | [Description] | Happy Path | Automated | [Expected result] | [ ] |

## Coverage Summary

| Epic | FRs | Scenarios | Automated | Manual | Deferred |
|------|-----|-----------|-----------|--------|----------|
| EPIC-001 | 2 | 4 | 3 | 1 | 0 |
| **Total** | | | | | |

Traceability

Every test scenario MUST reference:

• FR-ID — which functional requirement it validates
• EPIC-ID — which epic it belongs to
• TC-ID — unique test case identifier (TC-001, TC-002, ...)

Rules

1. PRD required — do not generate a test plan without a PRD
2. Full coverage — every acceptance criterion must have at least one test scenario
3. Unique IDs — every test case gets a TC-NNN identifier
4. Category balance — do not generate only happy path tests; include edge cases and error handling
5. No implementation — this skill produces test PLANS, not test code (use test skill for code)
6. Security scenarios — always include at least one security test per epic that handles user input
7. Measurable pass criteria — every scenario must have a concrete expected result, not "works correctly"