fix: agentic workflow waits for in-progress CI

Author: pinodeca

.github/aw/actions-lock.json

@@ -0,0 +1,14 @@
+{
+  "entries": {
+    "actions/github-script@v8": {
+      "repo": "actions/github-script",
+      "version": "v8",
+      "sha": "ed597411d8f924073f98dfc5c65a23a2325f34cd"
+    },
+    "github/gh-aw/actions/setup@v0.50.1": {
+      "repo": "github/gh-aw/actions/setup",
+      "version": "v0.50.1",
+      "sha": "fad43e3c91a4e1d43e458f68e96574127934e7d1"
+    }
+  }
+}

.github/workflows/fix-pr-pass-ci.md

@@ -62,14 +62,20 @@ Because PG18 uses `continue-on-error`, the overall CI status can appear **green
    gh run list --branch "$BRANCH" --workflow ci.yml --limit 5 --json databaseId,status,conclusion
    ```
 
-3. Check **individual job conclusions** (not just the overall run conclusion):
+3. **If the latest CI run is still `in_progress`**, wait for it to finish before proceeding. Poll every 30 seconds:
+   ```
+   gh run watch <run-id> --exit-status || true
+   ```
+   This blocks until the run completes. The `|| true` ensures you continue even if the run fails (which is expected).
+
+4. Check **individual job conclusions** (not just the overall run conclusion):
    ```
    gh run view <run-id> --json jobs --jq '.jobs[] | {name, conclusion, status}'
    ```
 
-4. Classify each job as `success`, `failure`, or `skipped`.
+5. Classify each job as `success`, `failure`, or `skipped`.
 
-5. If ALL jobs succeeded (including PG18), call the `noop` safe output with a message that CI is already passing and stop.
+6. If ALL jobs succeeded (including PG18), call the `noop` safe output with a message that CI is already passing and stop.
 
 ### Phase 2: Fix PG17 Failures First