qa-scaffolder

QA Scaffolder — Artifact Generator

Generate the complete QA infrastructure for a project's .claude/ directory using the scanner report and template files.

When This Skill Is Invoked

Called by /qa:init after the qa-scanner produces its report. Receives:

• The scanner report (project structure, modules, testing opportunities)
• The stack detection output (language, framework, test runner)
• The test analysis output (existing patterns, conventions)
• The selected test types (from user confirmation in init Step 4)

Workflow

Phase 1: Read Templates

Read template files from the plugin's templates/ directory based on what needs to be generated.

Always read (base templates):

• templates/skills/use-case-identifier.md
• templates/skills/scenario-writer.md
• templates/skills/coverage-analyzer.md
• templates/skills/test-reviewer.md
• templates/agents/test-runner.md
• templates/agents/coverage-checker.md
• templates/agents/test-reviewer.md
• templates/commands/test.md
• templates/commands/test-coverage.md
• templates/commands/test-review.md

Read per selected test type:

• Unit selected → templates/skills/test-writer.md
• Integration selected → templates/skills/integration-test-writer.md
• E2E selected → templates/skills/e2e-test-writer.md
• API selected → templates/skills/api-test-writer.md

Phase 2: Customize Templates

For each template, substitute project-specific details:

Variables to replace:

• {{language}} — detected language (e.g., "kotlin")
• {{framework}} — detected framework (e.g., "spring")
• {{testRunner}} — detected test runner (e.g., "junit")
• {{testCommand}} — detected unit test command (e.g., "gradle test")
• {{integrationTestCommand}} — integration test command (e.g., "gradle integrationTest")
• {{e2eTestCommand}} — e2e test command (e.g., "npx playwright test")
• {{apiTestCommand}} — API test command (e.g., "gradle test --tests '*ApiTest'")
• {{e2eRunner}} — e2e framework (e.g., "playwright", "cypress", "selenium")
• {{coverageTool}} — detected coverage tool (e.g., "jacoco")
• {{packageManager}} — detected package manager (e.g., "gradle")
• {{testLocation}} — test location convention (e.g., "src/test/")
• {{namingPattern}} — test naming pattern (e.g., "*Test.kt")
• {{modules}} — list of project modules from scanner
• {{existingPatterns}} — existing test patterns (if any)

Beyond simple substitution:

• Add project-specific examples based on actual module names
• Reference real file paths from the scanner report
• Include framework-specific testing patterns
• Adapt recommendations based on existing test conventions

Phase 3: Generate qa-config.json

{
  "stack": {
    "language": "{{language}}",
    "framework": "{{framework}}",
    "testRunner": "{{testRunner}}",
    "coverageTool": "{{coverageTool}}",
    "packageManager": "{{packageManager}}"
  },
  "testTypes": {
    "unit": {
      "enabled": true,
      "runner": "{{testRunner}}",
      "command": "{{testCommand}}"
    },
    "integration": {
      "enabled": false,
      "runner": null,
      "command": null
    },
    "e2e": {
      "enabled": false,
      "runner": null,
      "command": null
    },
    "api": {
      "enabled": false,
      "runner": null,
      "command": null
    }
  },
  "emphasis": {
    "unit": "standard",
    "integration": "standard",
    "e2e": "standard",
    "useCases": "standard"
  },
  "conventions": {
    "testLocation": "{{testLocation}}",
    "namingPattern": "{{namingPattern}}",
    "fixturePattern": "{{fixturePattern}}"
  },
  "coverage": {
    "tool": "{{coverageTool}}",
    "thresholds": { "lines": 80, "branches": 70 }
  },
  "generatedAt": "{{timestamp}}",
  "pluginVersion": "0.1.0"
}

Set enabled: true and populate runner + command only for test types the user selected in init Step 4.

Phase 4: Write Files

Write all generated files to the project's .claude/ directory:

Always written (base):

1. .claude/qa-config.json
2. .claude/skills/qa-use-case-identifier/SKILL.md
3. .claude/skills/qa-scenario-writer/SKILL.md
4. .claude/skills/qa-coverage-analyzer/SKILL.md
5. .claude/skills/qa-test-reviewer/SKILL.md
6. .claude/agents/qa-test-runner/AGENT.md
7. .claude/agents/qa-coverage-checker/AGENT.md
8. .claude/agents/qa-test-reviewer/AGENT.md
9. .claude/commands/test/COMMAND.md
10. .claude/commands/test-coverage/COMMAND.md
11. .claude/commands/test-review/COMMAND.md

Written per selected test type: 12. Unit → .claude/skills/qa-test-writer/SKILL.md 13. Integration → .claude/skills/qa-integration-test-writer/SKILL.md 14. E2E → .claude/skills/qa-e2e-test-writer/SKILL.md 15. API → .claude/skills/qa-api-test-writer/SKILL.md

Phase 5: Validate

Run verify-setup.js --project-root <path> to confirm all artifacts are valid.

If validation fails → fix the failing artifacts before completing.

Critical Rules

1. Specificity test — For each generated file, ask: "Could this have been generated without analyzing THIS project?" If yes → rewrite with more project-specific details.
2. Respect existing — If the project already has tests, the test-writer skills MUST reference existing conventions, not override them.
3. Complete files — Every generated file must be a complete, working artifact. No placeholders like "TODO" or "fill in later."
4. Ask before overwriting — If .claude/skills/qa-* files already exist, ask the user before overwriting.
5. Only generate what's selected — Do NOT generate test-writer skills for test types the user did not select.