@ngocsangairvds/vsaf 5.4.7 → 5.5.0

package/workflows/sdlc/master-sdlc-headless.yaml

@@ -1,73 +1,29 @@
 name: master-sdlc-headless
-description: Full SDLC pipeline (headless implement) — Discovery → PRD → Architecture → SRS → Test Design → Implement (headless) → Review → Feature Complete → Ship
+description: Full SDLC pipeline (headless implement, single-workspace) — Discovery → PRD → Architecture → SRS → Test Design → Implement (headless) → Review → Feature Complete → Ship
 nodes:
   - id: discovery
-    description: Phase — Discovery
-    bash: |
-      vsaf run sdlc/sdlc-discovery --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
+    workflow: sdlc/sdlc-discovery
   - id: prd
-    description: Phase — PRD
-    bash: |
-      vsaf run sdlc/sdlc-prd --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - discovery
+    workflow: sdlc/sdlc-prd
+    depends_on: [discovery]
   - id: architecture
-    description: Phase — Architecture
-    bash: |
-      vsaf run sdlc/sdlc-architecture --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - prd
+    workflow: sdlc/sdlc-architecture
+    depends_on: [prd]
   - id: srs
-    description: Phase — SRS
-    bash: |
-      vsaf run sdlc/sdlc-srs --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - architecture
+    workflow: sdlc/sdlc-srs
+    depends_on: [architecture]
   - id: test-design
-    description: Phase — Test Design
-    bash: |
-      vsaf run sdlc/sdlc-test-design --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - srs
+    workflow: sdlc/sdlc-test-design
+    depends_on: [srs]
   - id: implement
-    description: Phase — Implement (headless, engine-driven)
-    bash: |
-      vsaf run sdlc/sdlc-build-headless --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - test-design
+    workflow: sdlc/sdlc-build-headless
+    depends_on: [test-design]
   - id: review
-    description: Phase — Review
-    bash: |
-      vsaf run sdlc/sdlc-review --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - implement
+    workflow: sdlc/sdlc-review
+    depends_on: [implement]
   - id: complete
-    description: Phase — Feature Complete
-    bash: |
-      vsaf run sdlc/sdlc-feature-complete --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - review
+    workflow: sdlc/sdlc-feature-complete
+    depends_on: [review]
   - id: ship
-    description: Phase — Ship
-    bash: |
-      vsaf run sdlc/sdlc-ship --args $ARGUMENTS
-    workspace: false
-    timeout: 5400000
-    depends_on:
-      - complete
+    workflow: sdlc/sdlc-ship
+    depends_on: [complete]

package/workflows/sdlc/sdlc-build-headless.yaml

@@ -60,9 +60,15 @@ nodes:
       4. Write the chosen command to `$ARTIFACTS_DIR/run-tests.sh` — a POSIX sh script that runs the
          FULL test suite from the project root and exits with the test runner's exit code. Make it
          self-contained and re-runnable, then `chmod +x` it.
-      5. Verify once: `sh $ARTIFACTS_DIR/run-tests.sh`. It is FINE if tests fail now (the code isn't
-         written yet) — you are only proving the toolchain runs. Report one line: host or docker,
-         and that run-tests.sh is ready.
+      5. Verify once: `sh $ARTIFACTS_DIR/run-tests.sh 2>&1 | tee $ARTIFACTS_DIR/test-baseline.log`.
+         It is FINE if tests fail now (the code isn't written yet) — you are proving the toolchain
+         runs AND capturing the BASELINE (which tests already fail BEFORE this feature). From that
+         log, write `$ARTIFACTS_DIR/test-baseline.txt` with two lines exactly:
+           `total=<N>` (total tests the runner executed)
+           `failing=<comma-separated fully-qualified failing test names, or empty>`
+         (If the runner produced no test summary at all, write `total=0` and note the runner error —
+         that itself is a toolchain problem to fix here, not later.) Report one line: host or docker,
+         baseline total + failing count, and that run-tests.sh is ready.
 
   - id: impl-loop
     description: Implement 3/4 — TDD one case per turn; run tests IN-LOOP; finish only when GREEN
@@ -84,15 +90,22 @@ nodes:
          `<promise>IMPL_COMPLETE</promise>` and stop.
       3. For that one case: write/adjust its test (RED), implement the minimal code to make it pass
          (GREEN). Keep the change scoped to this case.
-      4. Run `sh $ARTIFACTS_DIR/run-tests.sh 2>&1 | tee $ARTIFACTS_DIR/test-output.log`. If ANY test
-         fails (this case OR a regression in a previously-passing test), FIX it now and re-run until
-         this case passes and nothing else is broken. Never move on with a failing suite.
+      4. Run the COMPLETE suite: `sh $ARTIFACTS_DIR/run-tests.sh 2>&1 | tee $ARTIFACTS_DIR/test-output.log`
+         (never a hand-picked subset). If ANY test fails (this case OR a regression in a
+         previously-passing test), FIX it now and re-run until this case passes and nothing else is
+         broken. Never move on with a failing suite.
       5. Tick the line to `- [x]` in `$ARTIFACTS_DIR/impl-progress.md` and append a one-line note.
 
       Work on the real project files. Match the project's existing structure, language, and test
-      framework. Do only the current case — the loop calls you again for the next. CRITICAL: NEVER
-      emit `<promise>IMPL_COMPLETE</promise>` while any test is failing; the signal means the whole
-      suite is green.
+      framework. Do only the current case — the loop calls you again for the next.
+
+      COMPLETION BAR — the full suite MUST be GREEN (0 failures, 0 errors) before you emit the
+      signal. A test that was ALREADY failing before this feature (see `$ARTIFACTS_DIR/test-baseline.txt`)
+      is NOT an excuse to ship red: either fix it, or — if it is genuinely unrelated and out of scope —
+      record it under a `## Pre-existing failures (quarantined)` section in impl-progress.md with the
+      exact test name + one-line rationale AND exclude it from the run so the suite is green. Do NOT
+      leave the suite red without an explicit quarantine entry. CRITICAL: NEVER emit
+      `<promise>IMPL_COMPLETE</promise>` while any test is failing; the signal means the whole suite is green.
     loop:
       until: IMPL_COMPLETE
       interactive: false
@@ -110,9 +123,20 @@ nodes:
     bash: |
       RT=$ARTIFACTS_DIR/run-tests.sh
       LOG=$ARTIFACTS_DIR/test-output.log
+      BASE=$ARTIFACTS_DIR/test-baseline.txt
       if [ ! -f "$RT" ]; then echo "[vsaf] test-gate: run-tests.sh missing (env-setup did not run)"; exit 1; fi
-      echo "[vsaf] test-gate: final verification via run-tests.sh"
+      echo "[vsaf] test-gate: final verification via run-tests.sh (FULL suite)"
       sh "$RT" > "$LOG" 2>&1; code=$?
       tail -40 "$LOG" 2>/dev/null || true
+      # Print the runner's real execution summary so the report is honest (never a subset count).
+      SUMMARY=$(grep -iE "Tests run: [0-9]+|[0-9]+ (passed|failed)|BUILD (SUCCESS|FAILURE)|failures=|[0-9]+ tests" "$LOG" 2>/dev/null | tail -6)
+      echo "[vsaf] test-gate summary:"; printf '%s\n' "$SUMMARY"
+      if [ -f "$BASE" ]; then echo "[vsaf] baseline (pre-feature): $(tr '\n' ' ' < "$BASE")"; fi
+      # Anti-false-green: refuse a 0 exit that shows NO evidence the suite actually ran
+      # (e.g. a runner that no-op'd or reused stale state). A green must be a proven green.
+      if [ "$code" -eq 0 ] && ! printf '%s' "$SUMMARY" | grep -qiE "Tests run: [0-9]+|[0-9]+ passed|[0-9]+ tests|BUILD SUCCESS"; then
+        echo "[vsaf] test-gate: FAIL — exit 0 but no test-run summary found; cannot confirm the suite executed. Refusing to false-green."
+        exit 1
+      fi
       echo "[vsaf] test-gate exit=$code"
       exit "$code"

package/workflows/sdlc/sdlc-design.yaml

@@ -35,27 +35,10 @@ nodes:
       - architecture-epics
     workspace: false
     command: sdlc-architecture-readiness
-  - id: srs-prereq
-    description: SRS 1/4 — input artifacts
+  - id: srs
+    description: Phase 4 — SRS specification, validation, interface verification (single cohesive skill)
+    model: sonnet
     workspace: false
-    command: sdlc-srs-prereq
+    command: sdlc-srs
     depends_on:
       - architecture-readiness
-  - id: srs-write
-    description: SRS 2/4 — write SRS
-    depends_on:
-      - srs-prereq
-    workspace: false
-    command: sdlc-srs-write
-  - id: srs-validate
-    description: SRS 3/4 — validate SRS
-    depends_on:
-      - srs-write
-    workspace: false
-    command: sdlc-srs-validate
-  - id: srs-interface
-    description: SRS 4/4 — interface verification + gate
-    depends_on:
-      - srs-validate
-    workspace: false
-    command: sdlc-srs-interface

package/workflows/sdlc/sdlc-review.yaml

@@ -28,9 +28,17 @@ nodes:
     workspace: false
     command: sdlc-review-impact
   - id: review-debate
-    description: Review 5/5 — debate + gate
+    description: Review 5/6 — debate + gate
     model: haiku
     depends_on:
       - review-impact
     workspace: false
     command: sdlc-review-debate
+  - id: review-fix
+    description: Review 6/6 — implement MUST-FIX findings, then re-verify the full suite
+    model: sonnet
+    permission_mode: bypassPermissions
+    depends_on:
+      - review-debate
+    workspace: false
+    command: sdlc-review-fix

package/workflows/sdlc/sdlc-srs.yaml

@@ -1,29 +1,8 @@
 name: sdlc-srs
-description: 'Phase workflow: SRS step-level execution'
+description: 'Phase workflow: SRS — single cohesive skill (write + 8-dimension validation + interface verification)'
 nodes:
-  - id: srs-prereq
-    description: SRS 1/4 — input artifacts
+  - id: srs
+    description: Phase 4 — SRS specification, validation, interface verification
     model: sonnet
     workspace: false
-    command: sdlc-srs-prereq
-  - id: srs-write
-    description: SRS 2/4 — write SRS
-    model: haiku
-    depends_on:
-      - srs-prereq
-    workspace: false
-    command: sdlc-srs-write
-  - id: srs-validate
-    description: SRS 3/4 — validate SRS
-    model: haiku
-    depends_on:
-      - srs-write
-    workspace: false
-    command: sdlc-srs-validate
-  - id: srs-interface
-    description: SRS 4/4 — interface verification + gate
-    model: haiku
-    depends_on:
-      - srs-validate
-    workspace: false
-    command: sdlc-srs-interface
+    command: sdlc-srs

package/skills/sdlc/srs/interface/SKILL.md

@@ -1,10 +0,0 @@
----
-name: sdlc-srs-interface
-description: "SRS 4/4 — interface verification + output"
-version: 1.0.0
-author: "@ngocsangairvds/vsaf"
----
-
-# SRS 4/4 — interface verification + output
-
-@include(step.md)

package/skills/sdlc/srs/interface/step.md

@@ -1,44 +0,0 @@
-### Step 4: Interface Verification
-
-```
-[SRS] [4/4] Interface verification via GitNexus... ⏳
-```
-
-Use GitNexus `shape_check` — verify code structure matches SRS assumptions.
-Append results to `05-srs.md` §Interface Contracts.
-
-```
-[SRS] [4/4] Interface verification via GitNexus... ✅
-```
-
-## Gate
-
-Human must approve SRS. Write `## Gate: APPROVED` at the end of the file.
-
-**Automated mode (opt-in):** when the environment has `VSAF_CI=1` or `VSAF_AUTO_APPROVE=1`, write the Gate yourself as `## Gate: APPROVED (auto · VSAF_CI · rationale: <one line>)` and proceed — the one-line rationale is REQUIRED for audit. Without the env flag, a human must approve (default).
-
-## Output
-
-- `.vsaf/docs/features/{feature-name}/05-srs.md` — SRS document (v4 template format)
-- `.vsaf/docs/features/{feature-name}/05-srs-validation-report.md` — Validation report with scores
-
-## Phase Exit Protocol
-
-In PHASE REPORT:
-- Output files: `05-srs.md`, `05-srs-validation-report.md`
-- Validation score: {overall_score}% ({GO/CONDITIONAL GO/NO-GO})
-- Suggestions: [1] Phase 5 Test Design (`/sdlc-test-design`) [2] Save artifacts
-
-Update `.vsaf/docs/STATUS.md`.
-
-**Notify the user:**
-> Artifacts have been saved. If the session is interrupted, open a new session and run the next phase — data is safe.
-
-## Rules
-
-- Every FR/NFR MUST have a unique traceability ID
-- DO NOT use vague language: "should", "might", "optionally" → replace with "MUST", "MUST NOT"
-- All terms MUST exist in CONTEXT.md
-- Every FR MUST be testable (have clear pass/fail criteria)
-- SRS MUST follow the v4 template structure (12 sections)
-- ⛔ DO NOT modify `graphify-out/` or `.gitnexus/`

package/skills/sdlc/srs/prereq/SKILL.md

@@ -1,10 +0,0 @@
----
-name: sdlc-srs-prereq
-description: "SRS 1/4 — input artifacts"
-version: 1.0.0
-author: "@ngocsangairvds/vsaf"
----
-
-# SRS 1/4 — input artifacts
-
-@include(step.md)

package/skills/sdlc/srs/prereq/step.md

@@ -1,19 +0,0 @@
-## Phase Entry Protocol
-
-```
-[SRS] [1/4] Checking input artifacts... ⏳
-```
-
-1. Input:
-   - `.vsaf/docs/features/{feature-name}/02-prd.md`
-   - `.vsaf/docs/features/{feature-name}/03-adr.md`
-   - `.vsaf/docs/features/{feature-name}/04-epics.md`
-2. Verify 02-prd.md AND 03-adr.md exist (04-epics.md optional):
-   - Any required file missing → Ask: "File not found. Have you run Phase 2 (PRD) and Phase 3 (Architecture)?"
-3. Read `.vsaf/docs/STATUS.md`
-4. Read `CONTEXT.md`
-5. Query `graph.json` for code structure + business mapping (if exists)
-
-```
-[SRS] [1/4] Checking input artifacts... ✅
-```

package/skills/sdlc/srs/validate/SKILL.md

@@ -1,10 +0,0 @@
----
-name: sdlc-srs-validate
-description: "SRS 3/4 — validate SRS"
-version: 1.0.0
-author: "@ngocsangairvds/vsaf"
----
-
-# SRS 3/4 — validate SRS
-
-@include(step.md)

package/skills/sdlc/srs/validate/step.md

@@ -1,23 +0,0 @@
-### Step 3: Validate SRS
-
-```
-[SRS] [3/4] Validating SRS (8 dimensions)... ⏳
-```
-
-Use `/sdlc-validate-srs` — validate SRS across 8 dimensions with weighted scoring.
-
-- Scope: single file (`05-srs.md`)
-- Cross-validate against: `02-prd.md`, `03-adr.md`
-- Output: `.vsaf/docs/features/{feature-name}/05-srs-validation-report.md`
-
-**Decision handling:**
-
-| Validation Result | Action |
-|---|---|
-| **GO** (≥80%, no CRITICAL) | Print score summary, proceed to Step 4 |
-| **CONDITIONAL GO** (60-79% or HIGH findings) | Print warnings + finding summary. Ask user: "Proceed despite warnings? (Y/N)" |
-| **NO-GO** (<60% or CRITICAL findings) | **STOP.** Print CRITICAL findings. Ask user to fix 05-srs.md and re-run `/sdlc-srs` |
-
-```
-[SRS] [3/4] Validating SRS (8 dimensions)... ✅
-```

package/skills/sdlc/srs/write/SKILL.md

@@ -1,10 +0,0 @@
----
-name: sdlc-srs-write
-description: "SRS 2/4 — write SRS"
-version: 1.0.0
-author: "@ngocsangairvds/vsaf"
----
-
-# SRS 2/4 — write SRS
-
-@include(step.md)

package/skills/sdlc/srs/write/step.md

@@ -1,19 +0,0 @@
-### Step 2: Write SRS
-
-```
-[SRS] [2/4] Writing SRS from PRD + ADR + Epics... ⏳
-```
-
-Use `/sdlc-write-srs` — write SRS following the v4 template (12 sections) from PRD + Architecture + Epics.
-
-Input:
-- `02-prd.md` as PRD source
-- `03-adr.md` as architecture context
-- `04-epics.md` as epic breakdown (optional)
-- Figma link from `CONTEXT.md` (optional)
-
-Output: `.vsaf/docs/features/{feature-name}/05-srs.md`
-
-```
-[SRS] [2/4] Writing SRS from PRD + ADR + Epics... ✅
-```