---

name: self-test-generator description: Generates tests directly from specifications and tasks. Use immediately after implementing a task. triggers: [task-complete, pre-review] outputs: [test-suite, coverage-report] depends_on: [no-guess-implementation]

Self Test Generator

Purpose

Generates behavior-based tests directly from specifications, not from code intuition. Tests are derived from what the spec says, not what the code does.

When to Use

• After completing an implementation task
• Before any manual review
• When spec changes require test updates

Generation Steps

1. Read Spec Sections: Identify all spec clauses relevant to the implemented task.
2. Map Spec to Test Cases: Generate happy path, edge case, error case, and constraint tests.
3. Write Behavior-Based Tests: Implement tests using relevant frameworks (JEST, PyTest, etc.).
4. Reference Spec in Tests: Every test MUST reference its source spec clause (e.g., // Spec: AUTH-001).

Decision Tree

Copyflowchart TD
    A[Start Test Gen] --> B{Spec ID found?}
    B -->|No| C[Stop - Refine Spec first]
    B -->|Yes| D{Multiple clauses?}
    D -->|Yes| E[Group tests by Spec ID]
    D -->|No| F[Create single test suite]
    E --> G{Complex logic?}
    F --> G
    G -->|Yes| H[Add extra edge case coverage]
    G -->|No| I[Happy path + basic error cases]
    H --> J[Verify Traceability]
    I --> J

Review Checklist

1. Traceability: Does every test link back to a valid Spec ID?
2. Isolation: Do tests avoid depending on implementation internals?
3. Completeness: Are edge cases and error states covered as per spec?
4. Determinism: Are tests reliable and independent of environment?

How to provide feedback

• Be specific: "The test for 'Login' fails to assert the 24h expiry requirement from AUTH-001."
• Explain why: "Without checking expiry, the test doesn't actually prove the spec is fully implemented."
• Suggest alternatives: "Recommend adding expect(token.expiry).toBe(86400) to the test."

Tests are spec enforcement, not code validation.