fix: separate stderr from stdout in benchmark to prevent invalid JSON (#1764)

* fix: separate stderr from stdout in benchmark to prevent invalid JSON

The benchmark run step used `2>&1` which mixed stderr progress logs into
benchmark-results.json, making downstream jq parsing silently fail.
Now stderr goes to benchmark-progress.log, and a JSON validation step
catches any corruption early.

Closes #1757

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: move artifact upload before JSON validation with if: always()

Address Copilot review: ensure artifacts are uploaded even when
validation fails, so logs are available for debugging.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: Mossaka

.github/workflows/performance-monitor.yml

@@ -47,16 +47,29 @@ jobs:
       - name: Run benchmarks
         id: benchmark
         run: |
-          npx tsx scripts/ci/benchmark-performance.ts > benchmark-results.json 2>&1 || true
+          npx tsx scripts/ci/benchmark-performance.ts > benchmark-results.json 2>benchmark-progress.log || true
+          cat benchmark-progress.log
+          echo "--- JSON output ---"
           cat benchmark-results.json
 
       - name: Upload results
+        if: always()
         uses: actions/upload-artifact@v4
         with:
           name: benchmark-results-${{ github.sha }}
-          path: benchmark-results.json
+          path: |
+            benchmark-results.json
+            benchmark-progress.log
           retention-days: 90
 
+      - name: Validate benchmark JSON
+        run: |
+          if ! jq empty benchmark-results.json 2>/dev/null; then
+            echo "ERROR: benchmark-results.json is not valid JSON"
+            cat benchmark-results.json
+            exit 1
+          fi
+
       - name: Check for regressions
         id: check
         run: |

scripts/ci/benchmark-performance.ts

@@ -9,6 +9,10 @@
  *  - Docker network creation time
  *
  * Outputs structured JSON with mean, median, p95, p99 per metric.
+ *
+ * IMPORTANT: stdout/stderr contract:
+ *   - stdout (console.log): JSON result only — consumed by the CI workflow via redirect to file
+ *   - stderr (console.error): progress messages and diagnostics — kept separate so JSON stays valid
  */
 
 import { execSync, ExecSyncOptions } from "child_process";