feat(ci): Separate action for loadtests

.github/actions/loadtest/action.yml

@@ -0,0 +1,256 @@
+name: 'Reloader Load Test'
+description: 'Run Reloader load tests with A/B comparison support'
+
+inputs:
+  old-ref:
+    description: 'Git ref for "old" version (optional, enables A/B comparison)'
+    required: false
+    default: ''
+  new-ref:
+    description: 'Git ref for "new" version (defaults to current checkout)'
+    required: false
+    default: ''
+  old-image:
+    description: 'Pre-built container image for "old" version (alternative to old-ref)'
+    required: false
+    default: ''
+  new-image:
+    description: 'Pre-built container image for "new" version (alternative to new-ref)'
+    required: false
+    default: ''
+  scenarios:
+    description: 'Scenarios to run: S1,S4,S6 or all'
+    required: false
+    default: 'S1,S4,S6'
+  test-type:
+    description: 'Test type label for summary: quick or full'
+    required: false
+    default: 'quick'
+  duration:
+    description: 'Test duration in seconds'
+    required: false
+    default: '60'
+  kind-cluster:
+    description: 'Name of existing Kind cluster (if empty, creates new one)'
+    required: false
+    default: ''
+  post-comment:
+    description: 'Post results as PR comment'
+    required: false
+    default: 'false'
+  pr-number:
+    description: 'PR number for commenting (required if post-comment is true)'
+    required: false
+    default: ''
+  github-token:
+    description: 'GitHub token for posting comments'
+    required: false
+    default: ${{ github.token }}
+  comment-header:
+    description: 'Optional header text for the comment'
+    required: false
+    default: ''
+
+outputs:
+  status:
+    description: 'Overall test status: pass or fail'
+    value: ${{ steps.run.outputs.status }}
+  summary:
+    description: 'Markdown summary of results'
+    value: ${{ steps.summary.outputs.summary }}
+  pass-count:
+    description: 'Number of passed scenarios'
+    value: ${{ steps.summary.outputs.pass_count }}
+  fail-count:
+    description: 'Number of failed scenarios'
+    value: ${{ steps.summary.outputs.fail_count }}
+
+runs:
+  using: 'composite'
+  steps:
+    - name: Determine images to use
+      id: images
+      shell: bash
+      run: |
+        # Determine old image
+        if [ -n "${{ inputs.old-image }}" ]; then
+          echo "old=${{ inputs.old-image }}" >> $GITHUB_OUTPUT
+        elif [ -n "${{ inputs.old-ref }}" ]; then
+          echo "old=localhost/reloader:old" >> $GITHUB_OUTPUT
+          echo "build_old=true" >> $GITHUB_OUTPUT
+        else
+          echo "old=" >> $GITHUB_OUTPUT
+        fi
+
+        # Determine new image
+        if [ -n "${{ inputs.new-image }}" ]; then
+          echo "new=${{ inputs.new-image }}" >> $GITHUB_OUTPUT
+        elif [ -n "${{ inputs.new-ref }}" ]; then
+          echo "new=localhost/reloader:new" >> $GITHUB_OUTPUT
+          echo "build_new=true" >> $GITHUB_OUTPUT
+        else
+          # Default: build from current checkout
+          echo "new=localhost/reloader:new" >> $GITHUB_OUTPUT
+          echo "build_new_current=true" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Build old image from ref
+      if: steps.images.outputs.build_old == 'true'
+      shell: bash
+      run: |
+        CURRENT_SHA=$(git rev-parse HEAD)
+        git checkout ${{ inputs.old-ref }}
+        docker build -t localhost/reloader:old .
+        echo "Built old image from ref: ${{ inputs.old-ref }}"
+        git checkout $CURRENT_SHA
+
+    - name: Build new image from ref
+      if: steps.images.outputs.build_new == 'true'
+      shell: bash
+      run: |
+        CURRENT_SHA=$(git rev-parse HEAD)
+        git checkout ${{ inputs.new-ref }}
+        docker build -t localhost/reloader:new .
+        echo "Built new image from ref: ${{ inputs.new-ref }}"
+        git checkout $CURRENT_SHA
+
+    - name: Build new image from current checkout
+      if: steps.images.outputs.build_new_current == 'true'
+      shell: bash
+      run: |
+        docker build -t localhost/reloader:new .
+        echo "Built new image from current checkout"
+
+    - name: Build loadtest binary
+      shell: bash
+      run: |
+        cd ${{ github.workspace }}/test/loadtest
+        go build -o loadtest ./cmd/loadtest
+
+    - name: Determine cluster name
+      id: cluster
+      shell: bash
+      run: |
+        if [ -n "${{ inputs.kind-cluster }}" ]; then
+          echo "name=${{ inputs.kind-cluster }}" >> $GITHUB_OUTPUT
+          echo "skip=true" >> $GITHUB_OUTPUT
+        else
+          echo "name=reloader-loadtest" >> $GITHUB_OUTPUT
+          echo "skip=false" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Load images into Kind
+      shell: bash
+      run: |
+        CLUSTER="${{ steps.cluster.outputs.name }}"
+
+        if [ -n "${{ steps.images.outputs.old }}" ]; then
+          echo "Loading old image: ${{ steps.images.outputs.old }}"
+          kind load docker-image "${{ steps.images.outputs.old }}" --name "$CLUSTER" || true
+        fi
+
+        echo "Loading new image: ${{ steps.images.outputs.new }}"
+        kind load docker-image "${{ steps.images.outputs.new }}" --name "$CLUSTER" || true
+
+    - name: Run load tests
+      id: run
+      shell: bash
+      run: |
+        cd ${{ github.workspace }}/test/loadtest
+
+        ARGS="--new-image=${{ steps.images.outputs.new }}"
+        ARGS="$ARGS --scenario=${{ inputs.scenarios }}"
+        ARGS="$ARGS --duration=${{ inputs.duration }}"
+        ARGS="$ARGS --cluster-name=${{ steps.cluster.outputs.name }}"
+
+        if [ -n "${{ steps.images.outputs.old }}" ]; then
+          ARGS="$ARGS --old-image=${{ steps.images.outputs.old }}"
+        fi
+
+        if [ "${{ steps.cluster.outputs.skip }}" = "true" ]; then
+          ARGS="$ARGS --skip-cluster"
+        fi
+
+        echo "Running: ./loadtest run $ARGS"
+        if ./loadtest run $ARGS; then
+          echo "status=pass" >> $GITHUB_OUTPUT
+        else
+          echo "status=fail" >> $GITHUB_OUTPUT
+        fi
+
+    - name: Generate summary
+      id: summary
+      shell: bash
+      run: |
+        cd ${{ github.workspace }}/test/loadtest
+
+        # Generate markdown summary
+        ./loadtest summary \
+          --results-dir=./results \
+          --test-type=${{ inputs.test-type }} \
+          --format=markdown > summary.md 2>/dev/null || true
+
+        # Output to GitHub Step Summary
+        cat summary.md >> $GITHUB_STEP_SUMMARY
+
+        # Store summary for output (using heredoc for multiline)
+        {
+          echo 'summary<<EOF'
+          cat summary.md
+          echo 'EOF'
+        } >> $GITHUB_OUTPUT
+
+        # Get pass/fail counts from JSON
+        COUNTS=$(./loadtest summary --format=json 2>/dev/null | head -20 || echo '{}')
+        echo "pass_count=$(echo "$COUNTS" | grep -o '"pass_count": [0-9]*' | grep -o '[0-9]*' || echo 0)" >> $GITHUB_OUTPUT
+        echo "fail_count=$(echo "$COUNTS" | grep -o '"fail_count": [0-9]*' | grep -o '[0-9]*' || echo 0)" >> $GITHUB_OUTPUT
+
+    - name: Post PR comment
+      if: inputs.post-comment == 'true' && inputs.pr-number != ''
+      uses: actions/github-script@v7
+      with:
+        github-token: ${{ inputs.github-token }}
+        script: |
+          const fs = require('fs');
+          const summaryPath = '${{ github.workspace }}/test/loadtest/summary.md';
+          let summary = 'No results available';
+          try {
+            summary = fs.readFileSync(summaryPath, 'utf8');
+          } catch (e) {
+            console.log('Could not read summary file:', e.message);
+          }
+
+          const header = '${{ inputs.comment-header }}';
+          const status = '${{ steps.run.outputs.status }}';
+          const statusEmoji = status === 'pass' ? ':white_check_mark:' : ':x:';
+
+          const body = [
+            header ? header : `## ${statusEmoji} Load Test Results (${{ inputs.test-type }})`,
+            '',
+            summary,
+            '',
+            '---',
+            `**Artifacts:** [Download](https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }})`,
+          ].join('\n');
+
+          await github.rest.issues.createComment({
+            owner: context.repo.owner,
+            repo: context.repo.repo,
+            issue_number: ${{ inputs.pr-number }},
+            body: body
+          });
+
+    - name: Upload results
+      uses: actions/upload-artifact@v4
+      if: always()
+      with:
+        name: loadtest-${{ inputs.test-type }}-results
+        path: |
+          ${{ github.workspace }}/test/loadtest/results/
+        retention-days: 30
+
+    - name: Cleanup Kind cluster (only if we created it)
+      if: always() && steps.cluster.outputs.skip == 'false'
+      shell: bash
+      run: |
+        kind delete cluster --name ${{ steps.cluster.outputs.name }} || true

.github/workflows/loadtest.yml

@@ -1,4 +1,4 @@
-name: Load Test
+name: Load Test (Full)
 
 on:
   issue_comment:
@@ -10,6 +10,7 @@ permissions:
   issues: write
 
 jobs:
+  # Full load test suite triggered by /loadtest command
   loadtest:
     # Only run on PR comments with /loadtest command
     if: |
@@ -45,6 +46,12 @@ jobs:
             core.setOutput('base_sha', pr.data.base.sha);
             console.log(`PR #${context.issue.number}: ${pr.data.head.ref} -> ${pr.data.base.ref}`);
 
+      - name: Checkout PR branch
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ steps.pr.outputs.head_sha }}
+          fetch-depth: 0  # Full history for building from base ref
+
       - name: Set up Go
         uses: actions/setup-go@v5
         with:
@@ -66,151 +73,23 @@ jobs:
           chmod +x kubectl
           sudo mv kubectl /usr/local/bin/kubectl
 
-      # Build OLD image from base branch (e.g., main)
+      - name: Run full A/B comparison load test
-      - name: Checkout base branch (old)
-        uses: actions/checkout@v4
-        with:
-          ref: ${{ steps.pr.outputs.base_ref }}
-          path: old
-
-      - name: Build old image
-        run: |
-          cd old
-          docker build -t localhost/reloader:old -f Dockerfile .
-          echo "Built old image from ${{ steps.pr.outputs.base_ref }} (${{ steps.pr.outputs.base_sha }})"
-
-      # Build NEW image from PR branch
-      - name: Checkout PR branch (new)
-        uses: actions/checkout@v4
-        with:
-          ref: ${{ steps.pr.outputs.head_ref }}
-          path: new
-
-      - name: Build new image
-        run: |
-          cd new
-          docker build -t localhost/reloader:new -f Dockerfile .
-          echo "Built new image from ${{ steps.pr.outputs.head_ref }} (${{ steps.pr.outputs.head_sha }})"
-
-      # Build and run loadtest from PR branch
-      - name: Build loadtest tool
-        run: |
-          cd new/test/loadtest
-          go build -o loadtest ./cmd/loadtest
-
-      - name: Run A/B comparison load test
         id: loadtest
-        run: |
+        uses: ./.github/actions/loadtest
-          cd new/test/loadtest
-          ./loadtest run \
-            --old-image=localhost/reloader:old \
-            --new-image=localhost/reloader:new \
-            --scenario=all \
-            --duration=60 2>&1 | tee loadtest-output.txt
-          echo "exitcode=${PIPESTATUS[0]}" >> $GITHUB_OUTPUT
-
-      - name: Upload results
-        uses: actions/upload-artifact@v4
-        if: always()
         with:
-          name: loadtest-results
+          old-ref: ${{ steps.pr.outputs.base_sha }}
-          path: |
+          new-ref: ${{ steps.pr.outputs.head_sha }}
-            new/test/loadtest/results/
+          scenarios: 'all'
-            new/test/loadtest/loadtest-output.txt
+          test-type: 'full'
-          retention-days: 30
+          post-comment: 'true'
-
+          pr-number: ${{ github.event.issue.number }}
-      - name: Post results comment
+          comment-header: |
-        uses: actions/github-script@v7
+            ## Load Test Results (Full A/B Comparison)
-        if: always()
+            **Comparing:** `${{ steps.pr.outputs.base_ref }}` → `${{ steps.pr.outputs.head_ref }}`
-        with:
+            **Triggered by:** @${{ github.event.comment.user.login }}
-          script: |
-            const fs = require('fs');
-
-            let results = '';
-            const resultsDir = 'new/test/loadtest/results';
-
-            // Collect summary of all scenarios
-            let passCount = 0;
-            let failCount = 0;
-            const summaries = [];
-
-            if (fs.existsSync(resultsDir)) {
-              const scenarios = fs.readdirSync(resultsDir).sort();
-              for (const scenario of scenarios) {
-                const reportPath = `${resultsDir}/${scenario}/report.txt`;
-                if (fs.existsSync(reportPath)) {
-                  const report = fs.readFileSync(reportPath, 'utf8');
-
-                  // Extract status from report
-                  const statusMatch = report.match(/Status:\s+(PASS|FAIL)/);
-                  const status = statusMatch ? statusMatch[1] : 'UNKNOWN';
-
-                  if (status === 'PASS') passCount++;
-                  else failCount++;
-
-                  // Extract key metrics for summary
-                  const actionMatch = report.match(/action_total\s+[\d.]+\s+[\d.]+\s+[\d.]+/);
-                  const errorsMatch = report.match(/errors_total\s+[\d.]+\s+[\d.]+/);
-
-                  summaries.push(`| ${scenario} | ${status === 'PASS' ? '✅' : '❌'} ${status} |`);
-
-                  results += `\n<details>\n<summary>${status === 'PASS' ? '✅' : '❌'} ${scenario}</summary>\n\n\`\`\`\n${report}\n\`\`\`\n</details>\n`;
-                }
-              }
-            }
-
-            if (!results) {
-              // Read raw output if no reports
-              if (fs.existsSync('new/test/loadtest/loadtest-output.txt')) {
-                const output = fs.readFileSync('new/test/loadtest/loadtest-output.txt', 'utf8');
-                const maxLen = 60000;
-                results = output.length > maxLen
-                  ? output.substring(output.length - maxLen)
-                  : output;
-                results = `\`\`\`\n${results}\n\`\`\``;
-              } else {
-                results = 'No results available';
-              }
-            }
-
-            const overallStatus = failCount === 0 ? '✅ ALL PASSED' : `❌ ${failCount} FAILED`;
-
-            const body = [
-              `## Load Test Results ${overallStatus}`,
-              '',
-              `**Comparing:** \`${{ steps.pr.outputs.base_ref }}\` (old) vs \`${{ steps.pr.outputs.head_ref }}\` (new)`,
-              `**Old commit:** ${{ steps.pr.outputs.base_sha }}`,
-              `**New commit:** ${{ steps.pr.outputs.head_sha }}`,
-              `**Triggered by:** @${{ github.event.comment.user.login }}`,
-              '',
-              '### Summary',
-              '',
-              '| Scenario | Status |',
-              '|----------|--------|',
-              summaries.join('\n'),
-              '',
-              `**Total:** ${passCount} passed, ${failCount} failed`,
-              '',
-              '### Detailed Results',
-              '',
-              results,
-              '',
-              '<details>',
-              '<summary>📦 Download full results</summary>',
-              '',
-              `Artifacts are available in the [workflow run](https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}).`,
-              '</details>',
-            ].join('\n');
-
-            await github.rest.issues.createComment({
-              owner: context.repo.owner,
-              repo: context.repo.repo,
-              issue_number: context.issue.number,
-              body: body
-            });
 
       - name: Add success reaction
-        if: success()
+        if: steps.loadtest.outputs.status == 'pass'
         uses: actions/github-script@v7
         with:
           script: |
@@ -222,7 +101,7 @@ jobs:
             });
 
       - name: Add failure reaction
-        if: failure()
+        if: steps.loadtest.outputs.status == 'fail'
         uses: actions/github-script@v7
         with:
           script: |

.github/workflows/pull_request.yaml

@@ -35,6 +35,7 @@ jobs:
 
     permissions:
       contents: read
+      pull-requests: write
 
     runs-on: ubuntu-latest
     name: Build
@@ -109,6 +110,17 @@ jobs:
     - name: Test
       run: make test
 
+    - name: Run quick A/B load tests
+      uses: ./.github/actions/loadtest
+      with:
+        old-ref: ${{ github.event.pull_request.base.sha }}
+        # new-ref defaults to current checkout (PR branch)
+        scenarios: 'S1,S4,S6'
+        test-type: 'quick'
+        kind-cluster: 'kind'  # Use the existing cluster created above
+        post-comment: 'true'
+        pr-number: ${{ github.event.pull_request.number }}
+
     - name: Generate Tags
       id: generate_tag
       run: |

Makefile

@@ -169,3 +169,43 @@ yq-install:
 	@curl -sL $(YQ_DOWNLOAD_URL) -o $(YQ_BIN)
 	@chmod +x $(YQ_BIN)
 	@echo "yq $(YQ_VERSION) installed at $(YQ_BIN)"
+
+# =============================================================================
+# Load Testing
+# =============================================================================
+
+LOADTEST_BIN = test/loadtest/loadtest
+LOADTEST_OLD_IMAGE ?= localhost/reloader:old
+LOADTEST_NEW_IMAGE ?= localhost/reloader:new
+LOADTEST_DURATION ?= 60
+LOADTEST_SCENARIOS ?= all
+
+.PHONY: loadtest-build loadtest-quick loadtest-full loadtest loadtest-clean
+
+loadtest-build: ## Build loadtest binary
+	cd test/loadtest && $(GOCMD) build -o loadtest ./cmd/loadtest
+
+loadtest-quick: loadtest-build ## Run quick load tests (S1, S4, S6)
+	cd test/loadtest && ./loadtest run \
+		--old-image=$(LOADTEST_OLD_IMAGE) \
+		--new-image=$(LOADTEST_NEW_IMAGE) \
+		--scenario=S1,S4,S6 \
+		--duration=$(LOADTEST_DURATION)
+
+loadtest-full: loadtest-build ## Run full load test suite
+	cd test/loadtest && ./loadtest run \
+		--old-image=$(LOADTEST_OLD_IMAGE) \
+		--new-image=$(LOADTEST_NEW_IMAGE) \
+		--scenario=all \
+		--duration=$(LOADTEST_DURATION)
+
+loadtest: loadtest-build ## Run load tests with configurable scenarios (default: all)
+	cd test/loadtest && ./loadtest run \
+		--old-image=$(LOADTEST_OLD_IMAGE) \
+		--new-image=$(LOADTEST_NEW_IMAGE) \
+		--scenario=$(LOADTEST_SCENARIOS) \
+		--duration=$(LOADTEST_DURATION)
+
+loadtest-clean: ## Clean loadtest binary and results
+	rm -f $(LOADTEST_BIN)
+	rm -rf test/loadtest/results

test/loadtest/cmd/loadtest/main.go

@@ -30,6 +30,15 @@ const (
 	testNamespace = "reloader-test"
 )
 
+// OutputFormat defines the output format for reports.
+type OutputFormat string
+
+const (
+	OutputFormatText     OutputFormat = "text"
+	OutputFormatJSON     OutputFormat = "json"
+	OutputFormatMarkdown OutputFormat = "markdown"
+)
+
 // workerContext holds all resources for a single worker (cluster + prometheus).
 type workerContext struct {
 	id          int
@@ -47,6 +56,7 @@ type Config struct {
 	Scenario     string
 	Duration     int
 	SkipCluster  bool
+	ClusterName  string // Custom cluster name (default: reloader-loadtest)
 	ResultsDir   string
 	ManifestsDir string
 	Parallelism  int // Number of parallel clusters (1 = sequential)
@@ -64,6 +74,8 @@ func main() {
 		runCommand(os.Args[2:])
 	case "report":
 		reportCommand(os.Args[2:])
+	case "summary":
+		summaryCommand(os.Args[2:])
 	case "help", "--help", "-h":
 		printUsage()
 	default:
@@ -78,7 +90,8 @@ func printUsage() {
 
 Usage:
   loadtest run [options]     Run A/B comparison tests
-  loadtest report [options]  Generate comparison report
+  loadtest report [options]  Generate comparison report for a scenario
+  loadtest summary [options] Generate summary across all scenarios (for CI)
   loadtest help              Show this help
 
 Run Options:
@@ -87,13 +100,21 @@ Run Options:
   --scenario=ID         Test scenario: S1-S13 or "all" (default: all)
   --duration=SECONDS    Test duration in seconds (default: 60)
   --parallelism=N       Run N scenarios in parallel on N clusters (default: 1)
-  --skip-cluster        Skip kind cluster creation (use existing, only for parallelism=1)
+  --skip-cluster        Skip kind cluster creation (use existing)
+  --cluster-name=NAME   Kind cluster name (default: reloader-loadtest)
   --results-dir=DIR     Directory for results (default: ./results)
 
 Report Options:
   --scenario=ID         Scenario to report on (required)
   --results-dir=DIR     Directory containing results (default: ./results)
   --output=FILE         Output file (default: stdout)
+  --format=FORMAT       Output format: text, json, markdown (default: text)
+
+Summary Options:
+  --results-dir=DIR     Directory containing results (default: ./results)
+  --output=FILE         Output file (default: stdout)
+  --format=FORMAT       Output format: text, json, markdown (default: markdown)
+  --test-type=TYPE      Test type label: quick, full (default: full)
 
 Examples:
   # Compare two images
@@ -111,8 +132,14 @@ Examples:
   # Run all 13 scenarios in parallel (one cluster per scenario)
   loadtest run --new-image=localhost/reloader:test --parallelism=13
 
-  # Generate report
+  # Generate report for a scenario
   loadtest report --scenario=S2 --results-dir=./results
+
+  # Generate JSON report
+  loadtest report --scenario=S2 --format=json
+
+  # Generate markdown summary for CI
+  loadtest summary --results-dir=./results --format=markdown
 `)
 }
 
@@ -122,6 +149,7 @@ func parseArgs(args []string) Config {
 		Duration:    60,
 		ResultsDir:  "./results",
 		Parallelism: 1,
+		ClusterName: clusterName, // default
 	}
 
 	// Find manifests dir relative to executable or current dir
@@ -151,6 +179,8 @@ func parseArgs(args []string) Config {
 			}
 		case arg == "--skip-cluster":
 			cfg.SkipCluster = true
+		case strings.HasPrefix(arg, "--cluster-name="):
+			cfg.ClusterName = strings.TrimPrefix(arg, "--cluster-name=")
 		case strings.HasPrefix(arg, "--results-dir="):
 			cfg.ResultsDir = strings.TrimPrefix(arg, "--results-dir=")
 		case strings.HasPrefix(arg, "--manifests-dir="):
@@ -234,15 +264,15 @@ func runCommand(args []string) {
 func runSequential(ctx context.Context, cfg Config, scenariosToRun []string, runtime string, runOld, runNew, runBoth bool) {
 	// Create cluster manager
 	clusterMgr := cluster.NewManager(cluster.Config{
-		Name:             clusterName,
+		Name:             cfg.ClusterName,
 		ContainerRuntime: runtime,
 	})
 
 	// Create/verify cluster
 	if cfg.SkipCluster {
-		log.Println("Skipping cluster creation (using existing)")
+		log.Printf("Skipping cluster creation (using existing cluster: %s)", cfg.ClusterName)
 		if !clusterMgr.Exists() {
-			log.Fatalf("Cluster %s does not exist. Remove --skip-cluster to create it.", clusterName)
+			log.Fatalf("Cluster %s does not exist. Remove --skip-cluster to create it.", cfg.ClusterName)
 		}
 	} else {
 		log.Println("Creating kind cluster...")
@@ -781,6 +811,7 @@ func cleanupReloader(ctx context.Context, version string, kubeContext string) {
 
 func reportCommand(args []string) {
 	var scenarioID, resultsDir, outputFile string
+	format := OutputFormatText
 	resultsDir = "./results"
 
 	for _, arg := range args {
@@ -791,6 +822,8 @@ func reportCommand(args []string) {
 			resultsDir = strings.TrimPrefix(arg, "--results-dir=")
 		case strings.HasPrefix(arg, "--output="):
 			outputFile = strings.TrimPrefix(arg, "--output=")
+		case strings.HasPrefix(arg, "--format="):
+			format = OutputFormat(strings.TrimPrefix(arg, "--format="))
 		}
 	}
 
@@ -803,7 +836,15 @@ func reportCommand(args []string) {
 		log.Fatalf("Failed to generate report: %v", err)
 	}
 
-	output := renderScenarioReport(report)
+	var output string
+	switch format {
+	case OutputFormatJSON:
+		output = renderScenarioReportJSON(report)
+	case OutputFormatMarkdown:
+		output = renderScenarioReportMarkdown(report)
+	default:
+		output = renderScenarioReport(report)
+	}
 
 	if outputFile != "" {
 		if err := os.WriteFile(outputFile, []byte(output), 0644); err != nil {
@@ -1584,3 +1625,302 @@ func renderScenarioReport(report *ScenarioReport) string {
 
 	return sb.String()
 }
+
+// renderScenarioReportJSON renders a scenario report as JSON.
+func renderScenarioReportJSON(report *ScenarioReport) string {
+	data, err := json.MarshalIndent(report, "", "  ")
+	if err != nil {
+		return fmt.Sprintf(`{"error": "%s"}`, err.Error())
+	}
+	return string(data)
+}
+
+// renderScenarioReportMarkdown renders a scenario report as concise markdown.
+func renderScenarioReportMarkdown(report *ScenarioReport) string {
+	var sb strings.Builder
+
+	// Status emoji
+	emoji := "✅"
+	if report.OverallStatus != "PASS" {
+		emoji = "❌"
+	}
+
+	sb.WriteString(fmt.Sprintf("## %s %s: %s\n\n", emoji, report.Scenario, report.OverallStatus))
+
+	if report.TestDescription != "" {
+		sb.WriteString(fmt.Sprintf("> %s\n\n", report.TestDescription))
+	}
+
+	// Key metrics table
+	sb.WriteString("| Metric | Value | Expected | Status |\n")
+	sb.WriteString("|--------|------:|:--------:|:------:|\n")
+
+	// Show only key metrics
+	keyMetrics := []string{"action_total", "reload_executed_total", "errors_total", "reconcile_total"}
+	for _, name := range keyMetrics {
+		for _, c := range report.Comparisons {
+			if c.Name == name {
+				value := fmt.Sprintf("%.0f", c.NewValue)
+				expected := "-"
+				if c.Expected > 0 {
+					expected = fmt.Sprintf("%.0f", c.Expected)
+				}
+				status := "✅"
+				if c.Status == "fail" {
+					status = "❌"
+				} else if c.Status == "info" {
+					status = "ℹ️"
+				}
+				sb.WriteString(fmt.Sprintf("| %s | %s | %s | %s |\n", c.DisplayName, value, expected, status))
+				break
+			}
+		}
+	}
+
+	return sb.String()
+}
+
+// ============================================================================
+// SUMMARY COMMAND
+// ============================================================================
+
+// SummaryReport aggregates results from multiple scenarios.
+type SummaryReport struct {
+	Timestamp  time.Time         `json:"timestamp"`
+	TestType   string            `json:"test_type"`
+	PassCount  int               `json:"pass_count"`
+	FailCount  int               `json:"fail_count"`
+	TotalCount int               `json:"total_count"`
+	Scenarios  []ScenarioSummary `json:"scenarios"`
+}
+
+// ScenarioSummary provides a brief summary of a single scenario.
+type ScenarioSummary struct {
+	ID          string  `json:"id"`
+	Status      string  `json:"status"`
+	Description string  `json:"description"`
+	ActionTotal float64 `json:"action_total"`
+	ActionExp   float64 `json:"action_expected"`
+	ErrorsTotal float64 `json:"errors_total"`
+}
+
+func summaryCommand(args []string) {
+	var resultsDir, outputFile, testType string
+	format := OutputFormatMarkdown // Default to markdown for CI
+	resultsDir = "./results"
+	testType = "full"
+
+	for _, arg := range args {
+		switch {
+		case strings.HasPrefix(arg, "--results-dir="):
+			resultsDir = strings.TrimPrefix(arg, "--results-dir=")
+		case strings.HasPrefix(arg, "--output="):
+			outputFile = strings.TrimPrefix(arg, "--output=")
+		case strings.HasPrefix(arg, "--format="):
+			format = OutputFormat(strings.TrimPrefix(arg, "--format="))
+		case strings.HasPrefix(arg, "--test-type="):
+			testType = strings.TrimPrefix(arg, "--test-type=")
+		}
+	}
+
+	summary, err := generateSummaryReport(resultsDir, testType)
+	if err != nil {
+		log.Fatalf("Failed to generate summary: %v", err)
+	}
+
+	var output string
+	switch format {
+	case OutputFormatJSON:
+		output = renderSummaryJSON(summary)
+	case OutputFormatText:
+		output = renderSummaryText(summary)
+	default:
+		output = renderSummaryMarkdown(summary)
+	}
+
+	if outputFile != "" {
+		if err := os.WriteFile(outputFile, []byte(output), 0644); err != nil {
+			log.Fatalf("Failed to write output file: %v", err)
+		}
+		log.Printf("Summary written to %s", outputFile)
+	} else {
+		fmt.Print(output)
+	}
+
+	// Exit with non-zero status if any tests failed
+	if summary.FailCount > 0 {
+		os.Exit(1)
+	}
+}
+
+func generateSummaryReport(resultsDir, testType string) (*SummaryReport, error) {
+	summary := &SummaryReport{
+		Timestamp: time.Now(),
+		TestType:  testType,
+	}
+
+	// Find all scenario directories
+	entries, err := os.ReadDir(resultsDir)
+	if err != nil {
+		return nil, fmt.Errorf("failed to read results directory: %w", err)
+	}
+
+	for _, entry := range entries {
+		if !entry.IsDir() || !strings.HasPrefix(entry.Name(), "S") {
+			continue
+		}
+
+		scenarioID := entry.Name()
+		report, err := generateScenarioReport(scenarioID, resultsDir)
+		if err != nil {
+			log.Printf("Warning: failed to load scenario %s: %v", scenarioID, err)
+			continue
+		}
+
+		scenarioSummary := ScenarioSummary{
+			ID:          scenarioID,
+			Status:      report.OverallStatus,
+			Description: report.TestDescription,
+		}
+
+		// Extract key metrics
+		for _, c := range report.Comparisons {
+			switch c.Name {
+			case "action_total":
+				scenarioSummary.ActionTotal = c.NewValue
+				scenarioSummary.ActionExp = c.Expected
+			case "errors_total":
+				scenarioSummary.ErrorsTotal = c.NewValue
+			}
+		}
+
+		summary.Scenarios = append(summary.Scenarios, scenarioSummary)
+		summary.TotalCount++
+		if report.OverallStatus == "PASS" {
+			summary.PassCount++
+		} else {
+			summary.FailCount++
+		}
+	}
+
+	// Sort scenarios by ID
+	sort.Slice(summary.Scenarios, func(i, j int) bool {
+		return naturalSort(summary.Scenarios[i].ID, summary.Scenarios[j].ID)
+	})
+
+	return summary, nil
+}
+
+// naturalSort compares two scenario IDs (S1, S2, ..., S10, S11)
+func naturalSort(a, b string) bool {
+	var aNum, bNum int
+	fmt.Sscanf(a, "S%d", &aNum)
+	fmt.Sscanf(b, "S%d", &bNum)
+	return aNum < bNum
+}
+
+func renderSummaryJSON(summary *SummaryReport) string {
+	data, err := json.MarshalIndent(summary, "", "  ")
+	if err != nil {
+		return fmt.Sprintf(`{"error": "%s"}`, err.Error())
+	}
+	return string(data)
+}
+
+func renderSummaryText(summary *SummaryReport) string {
+	var sb strings.Builder
+
+	sb.WriteString("================================================================================\n")
+	sb.WriteString("                       LOAD TEST SUMMARY\n")
+	sb.WriteString("================================================================================\n\n")
+
+	passRate := 0
+	if summary.TotalCount > 0 {
+		passRate = summary.PassCount * 100 / summary.TotalCount
+	}
+
+	fmt.Fprintf(&sb, "Test Type: %s\n", summary.TestType)
+	fmt.Fprintf(&sb, "Results:   %d/%d passed (%d%%)\n\n", summary.PassCount, summary.TotalCount, passRate)
+
+	fmt.Fprintf(&sb, "%-6s %-8s %-45s %10s %8s\n", "ID", "Status", "Description", "Actions", "Errors")
+	fmt.Fprintf(&sb, "%-6s %-8s %-45s %10s %8s\n", "------", "--------", strings.Repeat("-", 45), "----------", "--------")
+
+	for _, s := range summary.Scenarios {
+		desc := s.Description
+		if len(desc) > 45 {
+			desc = desc[:42] + "..."
+		}
+		actions := fmt.Sprintf("%.0f", s.ActionTotal)
+		if s.ActionExp > 0 {
+			actions = fmt.Sprintf("%.0f/%.0f", s.ActionTotal, s.ActionExp)
+		}
+		fmt.Fprintf(&sb, "%-6s %-8s %-45s %10s %8.0f\n", s.ID, s.Status, desc, actions, s.ErrorsTotal)
+	}
+
+	sb.WriteString("\n================================================================================\n")
+	return sb.String()
+}
+
+func renderSummaryMarkdown(summary *SummaryReport) string {
+	var sb strings.Builder
+
+	// Overall status
+	emoji := "✅"
+	title := "ALL TESTS PASSED"
+	if summary.FailCount > 0 {
+		emoji = "❌"
+		title = fmt.Sprintf("%d TEST(S) FAILED", summary.FailCount)
+	} else if summary.TotalCount == 0 {
+		emoji = "⚠️"
+		title = "NO RESULTS"
+	}
+
+	sb.WriteString(fmt.Sprintf("## %s Load Test Results: %s\n\n", emoji, title))
+
+	// Test type note
+	if summary.TestType == "quick" {
+		sb.WriteString("> 🚀 **Quick Test** (S1, S4, S6) — Use `/loadtest` for full suite\n\n")
+	}
+
+	// Pass rate
+	passRate := 0
+	if summary.TotalCount > 0 {
+		passRate = summary.PassCount * 100 / summary.TotalCount
+	}
+	sb.WriteString(fmt.Sprintf("**%d/%d passed** (%d%%)\n\n", summary.PassCount, summary.TotalCount, passRate))
+
+	// Results table
+	sb.WriteString("| | Scenario | Description | Actions | Errors |\n")
+	sb.WriteString("|:-:|:--------:|-------------|:-------:|:------:|\n")
+
+	for _, s := range summary.Scenarios {
+		icon := "✅"
+		if s.Status != "PASS" {
+			icon = "❌"
+		}
+
+		// Truncate description
+		desc := s.Description
+		if len(desc) > 45 {
+			desc = desc[:42] + "..."
+		}
+
+		// Format actions
+		actions := fmt.Sprintf("%.0f", s.ActionTotal)
+		if s.ActionExp > 0 {
+			actions = fmt.Sprintf("%.0f/%.0f", s.ActionTotal, s.ActionExp)
+		}
+
+		// Format errors
+		errors := fmt.Sprintf("%.0f", s.ErrorsTotal)
+		if s.ErrorsTotal > 0 {
+			errors = fmt.Sprintf("⚠️ %.0f", s.ErrorsTotal)
+		}
+
+		sb.WriteString(fmt.Sprintf("| %s | **%s** | %s | %s | %s |\n", icon, s.ID, desc, actions, errors))
+	}
+
+	sb.WriteString("\n📦 **[Download detailed results](../artifacts)**\n")
+
+	return sb.String()
+}