Skip to content

Changelog

All notable changes to the V-Model Extension Pack are documented here. The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

v0.7.0 — Bridge Commands & Pre-v0.8.0 Stabilization — 2026-05-03

Added — Bridge Commands

  • /speckit.v-model.plan — wraps /speckit.plan; synthesises a V-Model-enriched plan.md (5 H2 sections in fixed order: Summary, Technical Context, Constitution Check, Project Structure, Complexity Tracking) with additive HTML-comment annotations from the V-Model artifact set; output validated against pinned plan-schema-v0.7.0.json.
  • /speckit.v-model.tasks — wraps /speckit.tasks; produces a TDD-ordered tasks.md (12 H2s preserved verbatim) with hazard-driven priority elevation and per-HAZ verification tasks; each task carries an Implements-directive header.
  • /speckit.v-model.implement — wraps /speckit.implement; runs the deterministic 12-step pipeline (setup → 8-stage gate → domain overlay → codegen → 4-level test gen → splice → hallucination guard → quality harness → reduced-enrichment fallback → commit annotation → structured summary → handoffs).

Added — Lifecycle Hooks

  • after_specify — wires the V-Model directory layout under each new feature folder.
  • after_tasks — re-runs /speckit.v-model.trace to refresh the trace matrix after /speckit.v-model.tasks.
  • before_implement — re-runs the 8-stage run-v-model-gate.sh as a fail-fast check.
  • after_implement — re-runs /speckit.v-model.trace to refresh the matrix.

All four hooks are optional: true in v0.7.0 and become mandatory under the upcoming compliance_mode: strict profile (EPIC-1, v0.8.0).

Added — Validators & Orchestrator

  • run-v-model-gate.sh/.ps1 — 8-stage orchestrator: status → domain → matrix → requirement-coverage → system-coverage → architecture-coverage → module-coverage → hazard-coverage.
  • validate-artifact-status.sh/.ps1 (MF-6) — approval-status gate; default **Status**: Approved per artifact, configurable via --required-status.
  • validate-domain-profile.sh/.ps1 (MF-7) — domain config validator; absent-config = SKIP, present-and-invalid = fatal. Accepts only iso_26262, do_178c, iec_62304. Companion file: v-model-config.yml.example.
  • validate-core-schema.sh/.ps1 (MF-4) — three-pass validator (existence → ordering → wedge rejection) for plan.md (5 H2s) and tasks.md (12 H2s); enforced via diff -u against pinned heading sequences.
  • validate-implements-ids.sh/.ps1 — adds --canonical / --scan / --changed-only. Hallucination guard now scans the repo root (not just the V-Model dir) and intersects with git diff so that src/ and tests/{unit,integration,system,acceptance}/ are in scope.
  • setup-{plan,tasks,implement}.sh/.ps1 (MF-1) — V-Model-aware wrappers around the upstream setup-* scripts; surface VMODEL_DIR to the bridge prompts.

Changed — Splicer Hardening (MF-5)

  • splice-managed-regions.sh/.ps1 — BEGIN/END id mismatch and duplicate-id detection (exit 2); new --region-from <regions-file> per-region payload mode; diff -u original spliced emitted on stderr on every successful run for audit-trail capture. Single-payload invocations remain byte-identical on stdout.

Changed — Build Matrix Hardening (MF-3)

  • build-matrix.sh — same-dir mktemp plus EXIT/INT/TERM trap. No more /tmp writes; concurrent matrix builds are now safe.

Changed — Prompt + Docs Alignment (MF-9 + MF-10)

  • commands/plan.md no longer advertises a direct Implement Plan handoff; the V-Model lifecycle is now plan → tasks → implement.
  • README adds the Compliance & Hybrid Modes section distinguishing the compliant end-to-end V-Model bridge from the hybrid mode (feeding a V-Model tasks.md to core /speckit.implement). Hybrid mode is documented as a prototyping-only escape hatch.

Changed — Domain Set Narrowed (semantic break from v0.6.0)

  • The valid domain: value set is now strictly {iso_26262, do_178c, iec_62304}. The pre-v0.7.0 industry-vernacular synonyms (automotive, medical, aerospace) are rejected by validate-domain-profile. The general overlay has been removed with no replacement; non-regulated repositories should omit the domain: key (or the entire v-model-config.yml). See Configuration — Migration.

Documentation

  • New guide: Bridge Commands — the end-to-end bridge workflow, the 8-stage gate, the splicer, and the compliant-vs-hybrid distinction.

Stats

Metric Before (v0.6.0) After (v0.7.0)
Commands 14 17
Lifecycle hooks 0 4
Gate stages 6 8
BATS tests 364 455
Pester tests 347 431
Structural evals 89 89
LLM evals 42 53
End-to-end (E2E) tests — 32 (newly disclosed)

Known Limitations (deferred to v0.8.0)

  • EPIC-1 compliance_mode: strict profile — flips the 4 lifecycle hooks from optional: true to mandatory.
  • EPIC-2 Sealed baseline — cryptographic seal of the canonical V-Model artifact set per release tag.
  • EPIC-3 Immutable-artifact protection — pre-receive hook rejecting post-freeze edits to canonical artifacts.
  • EPIC-4 Semantic Implements-directive validation — beyond ID-existence to ID-relevance.
  • EPIC-5 v-model verify aggregator — single-command SARIF/JSON summary across all 8 gate stages.
  • EPIC-6 Authorized managed regions — splicer ACL by region id.
  • EPIC-7 Hybrid-mode telemetry — count and report bypassed gates.
  • EPIC-8 Per-domain trace-matrix presets — narrower validators per regulatory regime.
  • EPIC-9 Plan/Tasks schema migration tooling — versioned schema upgrade path beyond v0.7.0.
  • EPIC-10 Multi-domain projects — per-feature domain: overlays in monorepos.
  • specs/007-bridge-commands/ cross-cutting modules ARCH-008 / ARCH-014 / ARCH-016 lack dedicated ITP coverage; the V-Model artifact set is frozen at the Phase B paradigm lock and remediation is deferred to v0.8.0. Running run-v-model-gate.sh specs/007-bridge-commands will report ⅞ inner validators PASS and validate-architecture-coverage FAIL. This is the only known coverage gap on the dogfood feature.

v0.6.0 — Foundation Hardening — 2026-04-23

Changed — Domain Overlay Architecture

  • Decoupled all domain-specific safety content from 14 base commands into commands/overlays/{iso_26262,do_178c,iec_62304}/ directories
  • Base commands now contain only best-practice standards (IEEE, ISO/IEC); safety-specific content (ASIL tables, DAL classifications, etc.) loaded at runtime via domain: in v-model-config.yml
  • Added domain: field to config-template.yml (values: empty, iso_26262, do_178c, iec_62304)
  • Added assembly protocol: domain loading step at the top of each command's System Prompt
  • Created 36 overlay files (12 commands × 3 domains) with _domain.yml manifests

Changed — ID Lifecycle Model

  • Extended formal ID lifecycle model to all 8 generative commands (previously only acceptance had partial support)
  • Added [DEPRECATED — Superseded by {ID}] and [DEPRECATED — Withdrawn: <reason>] syntax to all commands
  • Added suspect cascade: downstream items automatically marked [SUSPECT — Parent {ID} deprecated] when parent is deprecated
  • Each command now includes a "Lifecycle Rules" section governing evolution behavior

Changed — Standards Enrichment

  • Enriched all 11 base commands with explicit ## Governing Standards sections
  • Integrated ISO/IEC 25010:2023 quality characteristics into requirements (NFR section) and system-design output
  • Integrated IEEE 1012:2016 V&V gates into acceptance, system-test, integration-test, and unit-test output
  • Integrated ISO/IEC 42030:2019 architecture evaluation into architecture-design output
  • Total standards referenced: 26 (17 best-practice + 9 safety-specific, up from 17)

Changed — Dogfooding

  • Evolved V-Model artifact chains for features 002–005e through ID lifecycle model
  • Created complete V-Model chain for feature 006 (foundation infrastructure)
  • All 89 structural evals updated to validate new standards sections; 364 BATS tests still pass

Stats

Metric Before (v0.5.0) After (v0.6.0)
Commands 14 14
Domain overlays 0 36
Standards referenced 17 26
Structural evals 89 89
BATS tests 364 364
Standards sections in artifacts 0 11 commands × 2+ sections

v0.5.0 — 2026-04-06

Added — New Commands

  • hazard-analysis — ISO 14971/26262 Failure Mode and Effects Analysis (FMEA) with HAZ-NNN hazard identifiers, operational state awareness, severity × likelihood risk matrix, mitigation traceability to REQ/SYS IDs, and progressive deepening (append-only at architecture level)
    • hazard-analysis-template.md — FMEA table template with 10 columns
    • validate-hazard-coverage.sh / validate-hazard-coverage.ps1 — Three-dimensional deterministic validator: forward (SYS→HAZ), backward (HAZ→REQ/SYS), and operational state consistency checks with --partial and --json flags
    • Matrix H (Hazard Traceability) in traceability matrix — HAZ → Mitigation → Verification linkage
    • HAZ-NNN ID pattern in id_validator.py
  • impact-analysis — Deterministic change impact analysis that builds a dependency graph from all V-Model markdown artifacts and traverses it to identify suspect artifacts affected by a change
    • --downward (default), --upward, and --full bidirectional traversal modes
    • --json flag for CI integration (structured JSON with blast radius, suspect artifacts by level, re-validation order)
    • Multi-ID support, <2s for 500+ IDs across 10+ artifact files
    • impact-analysis.sh / impact-analysis.ps1 — Bash and PowerShell scripts with awk-based graph parser and BFS traversal
  • peer-review — AI-powered stateless linter for any V-Model artifact, evaluating against standards-based criteria (INCOSE, IEEE 1016/42010, ISO 29119, ISO 14971, DO-178C) and producing PRF-{ARTIFACT}-NNN findings with severity classifications (Critical, Major, Minor, Observation)
    • Stateless linting model: findings regenerated from scratch each run, like ESLint
    • peer-review-check.sh / Peer-Review-Check.ps1 — CI parser scripts with exit codes: 0 (clean), 1 (Critical/Major — blocks PR), 2 (Minor — warning)
  • test-results — 100% deterministic JUnit XML + Cobertura XML ingestor that updates the traceability matrix in-place, flipping ⬜ Untested to ✅ Passed / ❌ Failed / ⏭️ Skipped with Date, Commit SHA, and optional Coverage columns
    • parse_test_results.py — stdlib-only Python helper (xml.etree.ElementTree) with 5 modules
    • ingest-test-results.sh / Ingest-Test-Results.ps1 — Bash and PowerShell wrappers (1:1 parity)
    • Coverage mapping via coverage-map.yml or convention-based matching from module-design.md
  • audit-report — 100% deterministic release audit report builder that produces a point-in-time release-audit-report.md for regulatory submission
    • Artifact inventory, traceability matrix embedding, coverage analysis, hazard management summary
    • Anomaly detection with waiver cross-referencing via waivers.md (WAV-NNN entries)
    • Compliance gating: ✅ RELEASE READY / ⚠️ RELEASE CANDIDATE / ❌ NOT READY
    • build-audit-report.sh / Build-Audit-Report.ps1 — Bash and PowerShell scripts (1:1 parity)

Added — Release Enhancements

  • validate-level.sh / Validate-Level.ps1 — Dispatch wrapper that invokes the correct validator for any V-Model level with --json and --partial flag support
  • Agent definitions (.github/agents/) for all 14 commands — previously only 3 existed
  • Sample CI workflow template (examples/github-actions/v-model-validation.yml)
  • 56 V-Model specification documents promoted from Draft to Approved

Added — Test Infrastructure

  • Hazard analysis fixtures: minimal (5 HAZ), complex (12 HAZ), gaps, golden/automotive-adas (15 HAZ), golden/medical-device (12 HAZ)
  • Impact analysis fixtures: linear, diamond, disconnected — with 17 golden JSON outputs
  • Peer review fixtures: clean, critical-major, minor-only, mixed-severity, observations-only
  • Test results fixtures: 8 JUnit XML scenarios, 2 Cobertura XML, 3 matrix fixtures, 10 golden JSON outputs
  • Audit report fixtures: clean, waived, blocking, orphaned-waiver, missing-required — with golden outputs
  • Python structural validators: hazard_validators.py, impact_validators.py
  • DeepEval metric wrappers: StructuralHazardAnalysisMetric, StructuralImpactAnalysisMetric

Changed

  • build-matrix.sh / build-matrix.ps1 extended with Matrix H generation block
  • trace command updated for five-matrix output (A + B + C + D + H)
  • classify_id() in both Bash and PowerShell now maps ALL compound prefixes
  • extension.yml updated with all 5 new commands (14 total)
  • Documentation updated across README, compliance-guide, id-schema-guide, usage-examples, product-vision, v-model-overview, and CONTRIBUTING

Stats

Metric Before After
Commands 9 14
Bash scripts 7 13
PowerShell scripts 7 13
BATS tests 91 364
Pester tests 91 347
Structural evals 51 89
LLM evals 36 42
Agent definitions 3 14

v0.4.0 — 2026-02-22

Added

  • module-design command — 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-test command — ISO 29119-4 white-box unit test plans with five named techniques and Dependency & Mock Registries
  • validate-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
  • Module design and unit test fixtures across all scenario directories
  • Module-level validators and structural/E2E evaluations
  • MOD-NNN, UTP-NNN-X, UTS-NNN-X# ID patterns
  • docs/id-schema-guide.md — Comprehensive guide to the four-tier ID schema

Changed

  • Extension version bumped from 0.3.0 to 0.4.0
  • setup-v-model now detects module-design.md and unit-test.md; 8 symmetric require flags
  • build-matrix extended with Matrix D generation
  • trace command updated from triple-matrix to quadruple-matrix output (A + B + C + D)
  • Renamed validate-coverage → validate-requirement-coverage for consistent naming

Fixed

  • BATS test for validate-system-coverage partial mode now correctly expects exit 0
  • PowerShell validate-system-coverage.ps1 now supports partial mode when system-test.md is absent
  • PowerShell validate-system-coverage.ps1 handles empty files via null-coalescing
  • Minimal module-design fixture now includes typed function signatures

Stats

Commands: 7 → 9 · BATS: 67 → 91 · Pester: 67 → 91 · Structural evals: 37 → 51 · LLM evals: 26 → 36

v0.3.0 — 2026-02-21

Added

  • architecture-design command — IEEE 42010/Kruchten 4+1 architecture decomposition with Logical, Process, Interface, and Data Flow views
  • integration-test command — ISO 29119-4 integration testing with Interface Contract, Data Flow, Fault Injection, and Concurrency techniques
  • validate-architecture-coverage.sh / validate-architecture-coverage.ps1 — Deterministic ARCH→ITP→ITS bidirectional coverage validation with CROSS-CUTTING module support
  • Matrix C (Integration Verification) in traceability matrix
  • Architecture and integration test fixtures across all scenario directories
  • Architecture-level validators and structural/E2E evaluations
  • ARCH-NNN, ITP-NNN-X, ITS-NNN-X# ID patterns
  • CROSS-CUTTING module tag for infrastructure/utility architecture modules

Changed

  • Extension version bumped from 0.2.0 to 0.3.0
  • build-matrix extended with Matrix C generation
  • trace command updated from dual-matrix to triple-matrix output (A + B + C)
  • Test fixture directories consolidated to shared scenario pattern

Stats

Commands: 5 → 7 · BATS: 48 → 67 · Pester: 48 → 67 · Structural evals: 21 → 37 · LLM evals: 16 → 26

v0.2.0 — 2026-02-20

Added

  • system-design command — IEEE 1016-compliant system component decomposition with four mandatory design views (Decomposition, Dependency, Interface, Data Design)
  • system-test command — ISO 29119-compliant system test plans with named testing techniques and technical BDD scenarios
  • Extended trace command — Dual-matrix traceability output (Matrix A + Matrix B)
  • System-level golden examples for medical device (CBGMS) and automotive ADAS (AEB)
  • E2E evaluation harness (tests/evals/harness.py)
  • Templates and helper scripts for system design and system test
  • Structural evaluations in PR CI (26 deterministic tests)

Changed

  • Template validators now accept both template-style and golden-fixture-style sections
  • validate-requirement-coverage and build-matrix scripts extended for dual-matrix support
  • Evals workflow updated with E2E job

v0.1.0 — 2026-02-19

Added

  • Extension scaffold with extension.yml manifest (schema v1.0)
  • requirements command — IEEE 29148 / INCOSE 8-criteria quality validation with four requirement categories (Functional, Non-Functional, Interface, Constraint)
  • acceptance command — Three-tier Acceptance Test Plan with test cases (ATP-NNN-X), BDD scenarios (SCN-NNN-X#), deterministic coverage validation, and append-only incremental updates
  • trace command — Regulatory-grade Bidirectional Traceability Matrix with 4 pillars (Strict Bidirectionality, Orphan & Gap Analysis, Versioning & Baselines, Granular Execution State)
  • Output templates for requirements, acceptance plan, and traceability matrix
  • Helper scripts (Bash + PowerShell): setup-v-model, validate-requirement-coverage, build-matrix, diff-requirements
  • Extension configuration template (config-template.yml)
  • Documentation: v-model-overview.md, usage-examples.md, compliance-guide.md
  • after_tasks hook for automatic traceability matrix generation
  • Self-documenting three-tier ID schema: REQ-NNN → ATP-NNN-X → SCN-NNN-X#

See also