/generate-tests

Generate test cases BEFORE implementation:

Arguments: $ARGUMENTS

Core Principle

Test-driven development: Generate tests first, then implement to pass them. This provides clear acceptance criteria and prevents "marking complete without testing".

Phase 1: Load Feature

1. Parse feature ID from arguments
2. Read feature from .claude-harness/features/active.json (or .claude-harness/feature-list.json for v2.x compat)
3. Verify feature exists and is in pending/planning phase

Phase 2: Analyze Project

4. Read .claude-harness/memory/semantic/architecture.json for:

   • Test framework (jest, vitest, pytest, etc.)
   • Test directory structure
   • Existing test patterns

5. Read .claude-harness/memory/procedural/successes.json for:

   • Test patterns that worked before
   • File naming conventions

Phase 3: Generate Test Cases

6. Based on feature description, generate test cases:

   • Unit tests for core functionality
   • Integration tests for API/database
   • Edge cases and error handling

7. Create test file at .claude-harness/features/tests/{feature-id}.json:

   {
     "featureId": "feature-XXX",
     "generatedAt": "<ISO timestamp>",
     "framework": "jest|pytest|vitest",
     "cases": [
       {
         "id": "test-001",
         "type": "unit|integration|e2e",
         "description": "Should do X when Y",
         "file": "tests/path/to/test.ts",
         "status": "pending",
         "code": "test code here",
         "dependencies": []
       }
     ],
     "coverage": {
       "target": 80,
       "current": 0
     }
   }
   

Phase 4: Create Test Files

8. Write actual test files to the project:

   • Create test file(s) based on project conventions
   • Tests should FAIL initially (no implementation yet)

9. Run tests to confirm they fail (expected)

Phase 5: Update Feature

10. Update feature in active.json:
    • Set status to "needs_implementation"
    • Set phase to "test_generation"
    • Set tests.generated = true
    • Set tests.file = path to test spec
    • Set tests.total = number of test cases

Phase 6: Report

11. Report:
    • Number of test cases generated
    • Test file locations
    • Expected failures (normal - no implementation yet)
    • Next: Run /claude-harness:implement feature-XXX to implement