v0.4.0 — Module Design ↔ Unit Testing (Feature 004)
Fourth V-Model layer completing the full Requirements → Acceptance, System Design → System Test, Architecture Design → Integration Test, Module Design → Unit Test chain.
Added
module-designcommand — DO-178C/ISO 26262-compliant low-level module designs with four mandatory views (Algorithmic/Logic, State Machine, Internal Data Structures, Error Handling & Return Codes)unit-testcommand — ISO 29119-4 white-box unit test plans with five named techniques (Statement & Branch Coverage, Boundary Value Analysis, Equivalence Partitioning, State Transition Testing, Strict Isolation) and Dependency & Mock Registriesvalidate-module-coverage.sh/validate-module-coverage.ps1— Deterministic ARCH→MOD→UTP→UTS bidirectional coverage validation with EXTERNAL and CROSS-CUTTING module support- Matrix D (Unit Verification) in traceability matrix — ARCH → MOD → UTP → UTS with parent ARCH annotations
--require-module-design,--require-unit-testflags for setup-v-model (bash + PowerShell)- Module design and unit test fixtures across all scenario directories (minimal, complex, gaps, empty, golden)
- Module-level validators and MOD-NNN, UTP-NNN-X, UTS-NNN-X# ID patterns
- EXTERNAL and DERIVED MODULE tags for third-party and emergent module designs
- Module design and unit test LLM-as-judge quality metrics and E2E evaluation tests
docs/id-schema-guide.md— Comprehensive guide to the four-tier ID schema, lifecycle, and end-to-end traceability
Changed
- Renamed
validate-coverage→validate-requirement-coveragefor consistentvalidate-{design-level}-coveragenaming convention - trace.md updated from triple-matrix to quadruple-matrix output (A + B + C + D)
- Total commands: 7 → 9; BATS tests: 67 → 91; Pester tests: 67 → 91; Structural evals: 37 → 51; LLM-as-judge evals: 26 → 36; E2E evals: 24 → 32
Fixed
- PowerShell
validate-system-coverage.ps1partial mode and empty file handling - Minimal module-design fixture typed signatures for eval compliance
Stats
- 91 files changed, +14,664 lines
- 9 commands, 4 validators, 4 traceability matrices
- 119 evals passing (51 structural + 36 LLM-as-judge + 32 E2E)
Full Changelog: v0.3.0...v0.4.0