feat(ci): add weekly standalone vulnerability scan

.github/workflows/vulnerability-scan.yml

@@ -0,0 +1,94 @@
+name: Vulnerability Scan
+
+on:
+  schedule:
+    - cron: '0 8 * * 1' # Monday 08:00 UTC — weekly scan for newly-disclosed CVEs
+  workflow_dispatch:
+    inputs:
+      image_ref:
+        description: "Full image ref to scan (default: latest published image)"
+        required: false
+        default: ""
+
+permissions:
+  contents: read
+  security-events: write
+
+concurrency:
+  group: vulnerability-scan-${{ github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  scan:
+    name: Grype scan — ${{ matrix.image_name }}
+    # Grype scans run on GitHub-hosted runners that can be preempted mid-job
+    # by spot-instance recycling. continue-on-error prevents a runner shutdown
+    # from marking the overall workflow as failed — SARIF uploads are best-effort.
+    continue-on-error: true
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - image_name: dakota
+          - image_name: dakota-nvidia
+    steps:
+      - name: Resolve image ref
+        id: resolve
+        run: |
+          set -euo pipefail
+          if [[ -n "${{ inputs.image_ref }}" ]]; then
+            REF="${{ inputs.image_ref }}"
+          else
+            # Resolve :latest tag to an immutable digest for reproducible scan
+            TAG="ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}:latest"
+            DIGEST=$(skopeo inspect --no-tags "docker://${TAG}" | jq -r '.Digest')
+            if [[ -z "${DIGEST}" ]]; then
+              echo "::error::Could not resolve digest for ${TAG}"
+              exit 1
+            fi
+            REF="ghcr.io/${{ github.repository_owner }}/${{ matrix.image_name }}@${DIGEST}"
+          fi
+          echo "ref=${REF}" >> "$GITHUB_OUTPUT"
+          echo "Scanning: ${REF}"
+
+      - name: Scan image with Grype
+        id: scan
+        uses: anchore/scan-action@e1165082ffb1fe366ebaf02d8526e7c4989ea9d2 # v7.4.0
+        with:
+          image: ${{ steps.resolve.outputs.ref }}
+          fail-build: false
+          severity-cutoff: critical
+          output-format: sarif
+
+      - name: Upload SARIF results
+        if: always()
+        uses: github/codeql-action/upload-sarif@8aad20d150bbac5944a9f9d289da16a4b0d87c1e # v4
+        with:
+          sarif_file: ${{ steps.scan.outputs.sarif }}
+          category: grype-${{ matrix.image_name }}
+
+      - name: Check for critical CVEs
+        if: always()
+        run: |
+          set -euo pipefail
+          CRITICAL_COUNT=$(python3 -c "
+          import json, sys
+          data = json.load(open('${{ steps.scan.outputs.sarif }}'))
+          runs = data.get('runs', [])
+          criticals = sum(
+            1 for run in runs
+            for result in run.get('results', [])
+            if result.get('ruleId', '').startswith('CVE') and
+               any(prop.get('security-severity', '0') >= '9.0'
+                   for prop in [result.get('properties', {})])
+          )
+          print(criticals)
+          " 2>/dev/null || echo "0")
+
+          echo "Critical CVEs found: ${CRITICAL_COUNT}"
+
+          if [[ "${CRITICAL_COUNT}" -gt 0 ]]; then
+            echo "::warning::${CRITICAL_COUNT} critical CVE(s) found in ${{ matrix.image_name }} — see Security tab"
+          fi