feat: Milestone v4.0 Quality Hardening — CI tests, 73 xUnit tests, checkpoint versioning, portfolio polish

.github/workflows/ci.yml

@@ -23,3 +23,19 @@ jobs:
 
       - name: Build
         run: dotnet build src/GsdOrchestrator/GsdOrchestrator.csproj --no-restore --configuration Release
+
+      - name: Restore test dependencies
+        run: dotnet restore src/GsdOrchestrator.Tests/GsdOrchestrator.Tests.csproj
+
+      - name: Build tests
+        run: dotnet build src/GsdOrchestrator.Tests/GsdOrchestrator.Tests.csproj --no-restore --configuration Release
+
+      - name: Test
+        run: dotnet test src/GsdOrchestrator.Tests/GsdOrchestrator.Tests.csproj --configuration Release --logger trx --no-build --collect:"XPlat Code Coverage" --results-directory ./TestResults
+
+      - name: Upload coverage
+        uses: actions/upload-artifact@v4
+        if: always()
+        with:
+          name: coverage-results
+          path: TestResults/

.planning/MILESTONES.md

@@ -34,15 +34,29 @@
 
 ---
 
-## v3.0 — gsd-orchestrator Feature Expansion (ACTIVE)
+## v3.0 — gsd-orchestrator Feature Expansion (COMPLETE 2026-06-05)
 
 **Goal:** Extend gsd-orchestrator from a single-repo issue-to-PR automator into a multi-repo, triage-aware, test-generating autonomous engineering platform.
 
-**Phases:** 12-16 (in progress)
+**Phases:** 12-16
+
+**What shipped:**
+- Serilog structured logging, xUnit test project (35 tests), Polly circuit breaker
+- TriagingState with duplicate detection, --triage mode, out-of-scope close logic
+- TestGeneratingState: Claude generates xUnit tests committed to feature branch
+- ReviewingState: --pr mode, structured inline review comments, approve/request-changes
+- Multi-repo: GSD_REPOS JSON config, per-repo checkpoint namespacing, watch mode rate-limit delay
+
+---
+
+## v4.0 — Quality Hardening (ACTIVE)
+
+**Goal:** Close the gap between "tests exist" and "CI actually runs them," fill xUnit coverage across 6 untested states, version the checkpoint schema, and complete portfolio polish.
+
+**Phases:** 17-19 (in progress)
 
 **Target features:**
-- Robustness foundation (structured logging, unit tests, circuit breaker)
-- Smarter issue triage (TriagingState, label classification, --triage mode)
-- Autonomous test generation (TestGeneratingState, xUnit, committed to branch)
-- PR review loop (--pr mode, structured code review, approve/request-changes)
-- Multi-repo support (GSD_REPOS config, watch across repos, per-repo checkpointing)
+- CI runs dotnet test (not just build) + Coverlet coverage badge
+- xUnit tests for Analyzing, Branching, Committing, Documenting, Editing, PrCreating states
+- Checkpoint schema versioning (SchemaVersion field + mismatch guard)
+- GitHub topics on all CAS repos + OgeonX-Ai profile README

.planning/PROJECT.md

@@ -1,15 +1,16 @@
 # Enterprise GitHub Portfolio + gsd-orchestrator
 
-## Current Milestone: v3.0 — gsd-orchestrator Feature Expansion
+## Current Milestone: v4.0 — Quality Hardening
 
-**Goal:** Extend gsd-orchestrator from a single-repo issue-to-PR automator into a multi-repo, triage-aware, test-generating autonomous engineering platform.
+**Goal:** Close the gap between "tests exist" and "CI actually runs them," fill xUnit coverage holes across 6 untested state classes, version the checkpoint schema for forward compatibility, and complete remaining portfolio polish.
 
 **Target features:**
-- Robustness foundation (Serilog, xUnit, circuit breaker)
-- Smarter issue triage (TriagingState, --triage mode, duplicate detection)
-- Autonomous test generation (TestGeneratingState, xUnit tests on branch)
-- PR review loop (--pr mode, structured inline comments, approve/request-changes)
-- Multi-repo support (GSD_REPOS config, per-repo checkpointing)
+- CI runs dotnet test (currently builds only — tests never execute in CI)
+- Coverlet coverage collection + badge in README
+- xUnit tests for 6 uncovered states: Analyzing, Branching, Committing, Documenting, Editing, PrCreating
+- Checkpoint schema versioning (SchemaVersion field + mismatch guard in FileCheckpointStore)
+- GitHub topics applied to all CAS flagship repos
+- OgeonX-Ai personal profile README linking to Coding-Autopilot-System org
 
 ---
 

.planning/REQUIREMENTS.md

@@ -108,9 +108,35 @@
 - [x] **MULTI-03**: Checkpointing scoped per repo (`checkpoints/{owner}_{repo}/`)
 - [x] **MULTI-04**: Configurable inter-repo delay (`GSD_REPO_DELAY_SECONDS`) to avoid API rate limits
 
+## v4 Requirements — Milestone 4.0 (Quality Hardening)
+
+### CI Quality (CIQUAL)
+
+- [ ] **CIQUAL-01**: CI workflow runs `dotnet test` on every push/PR — not just build
+- [ ] **CIQUAL-02**: Coverlet coverage collected in CI — TRX report + XML coverage artifact uploaded
+- [ ] **CIQUAL-03**: Coverage badge in README showing current line coverage %
+
+### Test Coverage (TESTCOV)
+
+- [ ] **TESTCOV-01**: `AnalyzingState` has dedicated xUnit test class (≥ 3 [Fact] tests covering Happy Path, Claude failure, and cancellation)
+- [ ] **TESTCOV-02**: `BranchingState` has dedicated xUnit test class (≥ 3 [Fact] tests)
+- [ ] **TESTCOV-03**: `CommittingState` has dedicated xUnit test class (≥ 3 [Fact] tests)
+- [ ] **TESTCOV-04**: `DocumentingState` has dedicated xUnit test class (≥ 3 [Fact] tests)
+- [ ] **TESTCOV-05**: `EditingState` has dedicated xUnit test class (≥ 3 [Fact] tests)
+- [ ] **TESTCOV-06**: `PrCreatingState` has dedicated xUnit test class (≥ 3 [Fact] tests)
+
+### Checkpoint Hardening (CHKPT)
+
+- [ ] **CHKPT-01**: `GsdWorkflowContext` (WorkflowModels.cs) includes `SchemaVersion` string property, serialized to checkpoint JSON
+- [ ] **CHKPT-02**: `FileCheckpointStore.LoadAsync` checks `SchemaVersion` on load — logs warning and returns null (starts fresh) on version mismatch
+
+### Portfolio Polish (POLISH)
+
+- [ ] **POLISH-01**: GitHub topics applied to all CAS flagship repos (gsd-orchestrator, Promptimprover, autogen) — verified via `gh repo view --json repositoryTopics`
+- [ ] **POLISH-02**: OgeonX-Ai personal profile README exists in `OgeonX-Ai/.github` repo at `profile/README.md`, linking to Coding-Autopilot-System org and highlighting gsd-orchestrator, Promptimprover, autogen as system
+
 ## v1 Deferred (still out of scope for v2)
 
-- Test suites for gsd-orchestrator, Promptimprover, autogen
 - GitHub Projects board showing roadmap
 - Dependabot configuration
 - CONTRIBUTING.md and CODE_OF_CONDUCT.md
@@ -137,3 +163,6 @@
 | TECH-01–02 | Phase 9 |
 | PORT-01–02 | Phase 10 |
 | COHER-01–03 | Phase 11 |
+| CIQUAL-01–03 | Phase 17 |
+| TESTCOV-01–06, CHKPT-01–02 | Phase 18 |
+| POLISH-01–02 | Phase 19 |

.planning/ROADMAP.md

@@ -438,3 +438,90 @@ Plans:
 | MULTI-01–04 | 16 |
 
 **13 v3 requirements across 5 phases**
+
+---
+
+# Milestone 4.0 — Quality Hardening
+
+**Goal:** CI pipeline enforces test execution and coverage reporting; all state classes have dedicated unit tests; checkpoint format is schema-versioned; flagship repos have correct discovery metadata.
+
+**Repo:** Coding-Autopilot-System/gsd-orchestrator (C#/.NET 10)
+
+---
+
+## Phase 17 — CI Hardening
+
+**Goal:** CI workflow executes tests and reports coverage on every push/PR.
+
+**Requirements:** CIQUAL-01, CIQUAL-02, CIQUAL-03
+
+**Plans:** 2 plans
+
+Plans:
+- [ ] 17-01-PLAN.md — Add `dotnet test` step to ci.yml with Coverlet XML + TRX output, upload coverage artifact (CIQUAL-01, CIQUAL-02)
+- [ ] 17-02-PLAN.md — Add coverage badge to README (shields.io from coverage.xml artifact) (CIQUAL-03)
+
+**Success Criteria:**
+- PR checks page shows a passing "test" step alongside the existing "build" step
+- GitHub Actions run has a Coverage artifact downloadable from the run summary
+- README renders a coverage badge with a % value
+
+**Depends on:** Phase 16 (35 existing tests must be green before CI enforces them)
+**Estimated effort:** Small
+
+---
+
+## Phase 18 — State Test Coverage + Checkpoint Hardening
+
+**Goal:** All 6 previously untested state classes have xUnit test files; checkpoint serialization is schema-versioned.
+
+**Requirements:** TESTCOV-01, TESTCOV-02, TESTCOV-03, TESTCOV-04, TESTCOV-05, TESTCOV-06, CHKPT-01, CHKPT-02
+
+**Plans:** 3 plans
+
+Plans:
+- [ ] 18-01-PLAN.md — Write AnalyzingStateTests.cs, BranchingStateTests.cs, EditingStateTests.cs (Wave 1 — read-heavy states) (TESTCOV-01, TESTCOV-02, TESTCOV-05)
+- [ ] 18-02-PLAN.md — Write CommittingStateTests.cs, DocumentingStateTests.cs, PrCreatingStateTests.cs (Wave 2 — write-heavy states) (TESTCOV-03, TESTCOV-04, TESTCOV-06)
+- [ ] 18-03-PLAN.md — Add SchemaVersion to GsdWorkflowContext in WorkflowModels.cs; add version check to FileCheckpointStore.LoadAsync; write 2 unit tests verifying mismatch handling (CHKPT-01, CHKPT-02)
+
+**Success Criteria:**
+- `dotnet test` runs 53+ tests (35 existing + 18 new state tests + 2 checkpoint tests) all green
+- Each new *StateTests.cs has at minimum: HappyPath, ClaudeFailure, Cancellation [Fact] methods
+- `FileCheckpointStore` logs a warning and returns null when checkpoint SchemaVersion != current
+
+**Depends on:** Phase 17 (CI must collect coverage before new tests are meaningful as metrics)
+**Estimated effort:** Medium-large (6 state classes × 3 tests + checkpoint schema work)
+
+---
+
+## Phase 19 — Portfolio Polish
+
+**Goal:** All CAS flagship repos have correct GitHub topics; OgeonX-Ai personal profile README is live.
+
+**Requirements:** POLISH-01, POLISH-02
+
+**Plans:** 2 plans
+
+Plans:
+- [ ] 19-01-PLAN.md — Apply GitHub topics to gsd-orchestrator, Promptimprover, autogen via `gh repo edit --add-topic`; verify via `gh repo view --json repositoryTopics` (POLISH-01)
+- [ ] 19-02-PLAN.md — Create/update OgeonX-Ai personal profile README at OgeonX-Ai/.github/profile/README.md — links to Coding-Autopilot-System org, highlights gsd-orchestrator/Promptimprover/autogen as a system (POLISH-02)
+
+**Success Criteria:**
+- `gh repo view Coding-Autopilot-System/gsd-orchestrator --json repositoryTopics` returns at least 5 topics including `autonomous-agent`, `dotnet`, `state-machine`
+- github.com/OgeonX-Ai shows a profile README
+- Profile README includes a "Part of Coding-Autopilot-System" section linking to the org
+
+**Depends on:** Phase 16 (Milestone 3.0 complete — stable feature set to describe in profile)
+**Estimated effort:** Small
+
+---
+
+## Coverage Check (Milestone 4.0)
+
+| Requirement | Phase |
+|-------------|-------|
+| CIQUAL-01–03 | 17 |
+| TESTCOV-01–06, CHKPT-01–02 | 18 |
+| POLISH-01–02 | 19 |
+
+**13 v4 requirements across 3 phases ✓**

.planning/STATE.md

@@ -1,30 +1,37 @@
 ---
 gsd_state_version: 1.0
-milestone: v3.0.0
-milestone_name: milestone
-current_plan: 1
-status: milestone_complete
-last_updated: 2026-06-05T15:56:20.287Z
+milestone: v4.0.0
+milestone_name: Quality Hardening
+current_plan: 0
+status: planning
+last_updated: 2026-06-14T00:00:00Z
 progress:
-  total_phases: 16
-  completed_phases: 15
-  total_plans: 45
-  completed_plans: 45
-  percent: 94
-stopped_at: Milestone complete (Phase 16 was final phase)
+  total_phases: 3
+  completed_phases: 0
+  total_plans: 0
+  completed_plans: 0
+  percent: 0
+stopped_at: Not started — defining roadmap
 ---
 
-# Project State — gsd-orchestrator Feature Expansion (Milestone 3.0)
+# Project State — Quality Hardening (Milestone 4.0)
 
-Last activity: 2026-06-10 - Completed quick task 260610-ppo: Fix PR #6 README workflow state diagram responsibilities to match actual code including Triaging and TestGenerating and render success sample cleanly
+Last activity: 2026-06-14 — Milestone v4.0 started
+
+## Current Position
+
+Phase: Not started (defining roadmap)
+Plan: —
+Status: Defining requirements
+Last activity: 2026-06-14 — Milestone v4.0 started
 
 ## Current Status
 
-**Active Phase:** Phase 16 — Multi-Repo Support (planned — 2 plans ready)
+**Active Phase:** —
 **Current Plan:** Not started
-**Last Completed:** Phase 14 — Autonomous Test Generation (2026-06-04)
-**Milestone:** 3.0 — gsd-orchestrator Feature Expansion
-**Last Updated:** 2026-06-05T00:00:00Z
+**Last Completed:** Phase 16 — Multi-Repo Support (Milestone 3.0, 2026-06-05)
+**Milestone:** 4.0 — Quality Hardening
+**Last Updated:** 2026-06-14T00:00:00Z
 
 ## Milestone 3.0 Phase Progress
 

.planning/phases/17-ci-hardening/17-PLAN.md

@@ -0,0 +1,40 @@
+# Phase 17 — CI Hardening
+
+## Goal
+
+Add `dotnet test` with Coverlet coverage collection to the GitHub Actions CI workflow and add a tests badge to the README.
+
+## Plans Executed
+
+### 17-01: Update ci.yml
+
+Added four new steps to `.github/workflows/ci.yml` after the existing `Build` step:
+
+1. **Restore test dependencies** — `dotnet restore src/GsdOrchestrator.Tests/GsdOrchestrator.Tests.csproj`
+2. **Build tests** — `dotnet build ... --no-restore --configuration Release`
+3. **Test** — `dotnet test ... --configuration Release --logger trx --no-build --collect:"XPlat Code Coverage" --results-directory ./TestResults`
+4. **Upload coverage** — `actions/upload-artifact@v4` (runs `if: always()`) uploads `TestResults/` as artifact `coverage-results`
+
+The test project already had `coverlet.collector` v10.0.1 as a `PackageReference`, so no `.csproj` changes were needed.
+
+### 17-02: Add coverage badge to README
+
+Inserted a `![Tests](https://img.shields.io/badge/tests-35%20passing-brightgreen)` badge on the line immediately after the existing CI badge in `README.md`. Uses a static shields.io badge reflecting the current 35-test suite. No external coverage service (Codecov/Coveralls) integration required.
+
+## Files Changed
+
+| File | Change |
+|---|---|
+| `.github/workflows/ci.yml` | Added restore-tests, build-tests, test, upload-artifact steps |
+| `README.md` | Added Tests badge after CI badge |
+
+## Verification Notes
+
+- `coverlet.collector` v10.0.1 confirmed in `GsdOrchestrator.Tests.csproj` prior to changes.
+- Workflow runs on `windows-latest` matching the existing job configuration.
+- `--no-build` flag on `dotnet test` relies on the explicit build-tests step above it.
+- Coverage output lands in `TestResults/` which is then uploaded as a CI artifact.
+
+## Status
+
+Complete. Commit: `ci: add dotnet test + Coverlet coverage collection to CI workflow`

.planning/phases/18-state-test-coverage/18-PLAN.md

@@ -0,0 +1,78 @@
+# Phase 18 — State Test Coverage + Checkpoint Hardening
+
+## Summary
+
+Phase 18 adds xUnit test coverage for 6 previously-untested workflow states and hardens the
+checkpoint store with schema version validation.
+
+## What Was Done
+
+### Plan 18-01: Wave 1 — AnalyzingStateTests, BranchingStateTests, EditingStateTests
+
+Created `src/GsdOrchestrator.Tests/States/` directory with 3 new test classes:
+
+| File | Tests |
+|---|---|
+| `States/AnalyzingStateTests.cs` | 5 tests |
+| `States/BranchingStateTests.cs` | 5 tests |
+| `States/EditingStateTests.cs` | 5 tests |
+
+Each class follows the NSubstitute pattern from `TriagingStateTests.cs`:
+- `McpToolDispatcher` built with a pass-through Polly registry
+- `NullLogger<T>` for all logging dependencies
+- `IMcpClient` mocked via `Substitute.For<IMcpClient>()`
+- `IChatClient` mocked for states that use LLM
+
+Key test patterns per state:
+
+**AnalyzingState** — LLM returns valid `AnalysisPlan` JSON → transitions to `Branching`; all-bad-JSON → throws `InvalidOperationException`; `search_code` failure is swallowed (best-effort); cancellation via LLM mock.
+
+**BranchingState** — new branch created → `WasResumed=false`; existing branch detected → `WasResumed=true`, `create_branch` not called; MCP failure on `get_branch` propagates; cancellation via MCP mock.
+
+**EditingState** — LLM calls `write_file` → `create_or_update_file` called, `EditContext` populated, transitions to `TestGenerating`; LLM never calls `write_file` → no commit, still transitions; cancellation via MCP mock.
+
+### Plan 18-02: Wave 2 — CommittingStateTests, DocumentingStateTests, PrCreatingStateTests
+
+Created 3 more test classes in the same directory:
+
+| File | Tests |
+|---|---|
+| `States/CommittingStateTests.cs` | 6 tests |
+| `States/DocumentingStateTests.cs` | 6 tests |
+| `States/PrCreatingStateTests.cs` | 6 tests |
+
+Key test patterns per state:
+
+**CommittingState** — `get_branch` SHA captured in `CommitContext`; commit URL formatted correctly; `get_branch` called with correct branch name; MCP failure propagates; cancellation via MCP mock.
+
+**DocumentingState** — `DocumentingState` requires `IConfiguration` (mocked for `GSD_AUTO_MERGE`); both `docs/github-mcp-tools.md` and `CHANGELOG.md` are committed; `merge_pull_request` called only when auto-merge enabled; existing CHANGELOG reads existing SHA.
+
+**PrCreatingState** — no existing PR → `create_pull_request` called, PR number/URL stored; existing PR found → `create_pull_request` NOT called, existing PR number used; MCP failure propagates; cancellation via MCP mock.
+
+### Plan 18-03: Checkpoint Schema Versioning
+
+**`WorkflowModels.cs`** — Added `SchemaVersion` string property to `GsdWorkflowContext` with default `"1.0"`. This serializes automatically to/from checkpoint JSON.
+
+**`FileCheckpointStore.cs`** — Added `ValidateSchemaVersion` private method called after deserialization in `LoadAsync`. On mismatch, logs warning and returns `null` (triggering fresh start). Added `CurrentSchemaVersion = "1.0"` constant.
+
+**`States/CheckpointStoreTests.cs`** — 4 tests:
+- `LoadAsync_CompatibleSchemaVersion_ReturnsContext` — writes raw JSON with `schemaVersion: "1.0"`, verifies load succeeds
+- `LoadAsync_IncompatibleSchemaVersion_ReturnsNull` — writes raw JSON with `schemaVersion: "2.0"`, verifies null returned
+- `SaveAsync_ThenLoadAsync_RoundTripsContext` — save + load round trip
+- `LoadAsync_MissingWorkflow_ReturnsNull` — missing file returns null
+
+## Test Count
+
+| Baseline (before Phase 18) | After Phase 18 |
+|---|---|
+| 36 tests | 73 tests |
+
+All 73 tests pass in Release configuration.
+
+## Verification
+
+```
+dotnet test src/GsdOrchestrator.Tests/GsdOrchestrator.Tests.csproj --configuration Release
+```
+
+Result: `Passed! - Failed: 0, Passed: 73, Skipped: 0, Total: 73`