feat: surface sustained runtime reliability evidence

README.md

@@ -127,6 +127,7 @@ The Lab decision surface now also exposes `policy_version`, `triggered_rules`, a
 
 `agent-runtime-report` is an additive reliable edge agent runtime report path.
 It bundles Orchestrator scheduling evidence and AIGuard runtime reliability `guard_analysis` into a Lab-owned agent deployment decision context without changing existing Runtime result or compare contracts.
+The report preserves sustained queue-depth, policy decision reason, and `sustained_overload_risk` evidence as local-first deployment review context.
 
 ![InferEdge Local Studio demo evidence](assets/images/local-studio-demo-evidence.png)
 

docs/portfolio/agent_runtime_reliability_report.md

@@ -45,6 +45,12 @@ poetry run inferedgelab agent-runtime-report \
 | fallback_rate | 0.583333 |
 | deadline_miss_rate | 0.1 |
 | queue_backlog_policy_decision_count | 1 |
+| max_total_queue_depth | 6 |
+| top_policy_decision_reason | queue_backlog_threshold_exceeded |
+
+AIGuard `guard_analysis` also includes `sustained_overload_risk`, which Lab
+preserves as report evidence and reflects in the agent deployment decision
+context.
 
 ## Lab Decision Context
 
@@ -67,6 +73,7 @@ Triggered rules:
 - `fallback_rate_block`
 - `deadline_miss_review`
 - `queue_backlog_review`
+- `sustained_overload_review`
 
 ## Boundary
 

examples/agent_runtime/agent_3_orchestration_summary.json

@@ -1,5 +1,10 @@
 {
   "schema_version": "inferedge-orchestration-summary-v1",
+  "run": {
+    "name": "agent_3_workload_sustained_high_load",
+    "scenario_mode": "sustained_high_load",
+    "frame_interval_ms": 5.0
+  },
   "agent_runtime_summary": {
     "schema_version": "inferedge-orchestration-summary-v1",
     "source_contracts": {
@@ -41,12 +46,53 @@
       "overload_event_count": 14
     }
   },
+  "sustained_runtime_summary": {
+    "schema_version": "inferedge-orchestrator-sustained-summary-v1",
+    "scenario_mode": "sustained_high_load",
+    "queue_depth_sample_count": 1,
+    "latency_sample_count": 1,
+    "max_total_queue_depth": 6,
+    "deadline_missed_count": 1,
+    "dropped_count": 14,
+    "fallback_count": 14,
+    "policy_decision_count": 14,
+    "overload_event_count": 14
+  },
+  "queue_depth_timeline": [
+    {
+      "cycle": 1,
+      "stage": "before_policy",
+      "queue_depth": {
+        "vision_agent": 4,
+        "voice_command_agent": 2,
+        "safety_monitor_agent": 0
+      },
+      "total_queue_depth": 6
+    }
+  ],
+  "latency_timeline": [
+    {
+      "agent_id": "vision_agent",
+      "task_id": "task_vision_agent",
+      "latency_ms": 41.0,
+      "latency_budget_ms": 33.0,
+      "deadline_missed": true
+    }
+  ],
   "policy_decision_log": [
     {
       "agent_id": "vision_agent",
       "task_id": "task_vision_agent",
       "decision": "load_shedding",
       "reason": "queue_backlog_threshold_exceeded",
+      "decision_reason": "queue_backlog_threshold_exceeded",
+      "total_backlog_before": 6,
+      "backlog_threshold": 3,
+      "queue_depth_snapshot": {
+        "vision_agent": 4,
+        "voice_command_agent": 2,
+        "safety_monitor_agent": 0
+      },
       "fallback_used": true,
       "protected_agent_id": "safety_monitor_agent"
     }

examples/agent_runtime/aiguard_runtime_guard_analysis.json

@@ -31,15 +31,41 @@
         "executed_count": 10,
         "dropped_count": 14
       }
+    },
+    {
+      "type": "sustained_overload_risk",
+      "metric_name": "max_total_queue_depth",
+      "observed_value": 6,
+      "baseline_value": null,
+      "threshold": 3,
+      "delta": null,
+      "delta_pct": null,
+      "increase_factor": null,
+      "severity": "medium",
+      "status": "failed",
+      "explanation": "Queue depth grew under sustained high-load mode.",
+      "why_it_matters": "Sustained queue depth growth shows that incoming workload pressure can exceed edge-device execution capacity.",
+      "suspected_causes": [
+        "sustained_multi_agent_overload",
+        "producer_rate_exceeds_scheduler_capacity"
+      ],
+      "recommendation": "Lower producer rate, tighten stale-frame drop policy, or move lower priority work behind a fallback path before deployment.",
+      "raw_context": {
+        "scenario_mode": "sustained_high_load",
+        "queue_depth_sample_count": 1,
+        "latency_sample_count": 1
+      }
     }
   ],
   "suspected_causes": [
     "queue_backlog",
     "overload_load_shedding",
-    "producer_rate_exceeds_runtime_capacity"
+    "producer_rate_exceeds_runtime_capacity",
+    "sustained_multi_agent_overload"
   ],
   "recommendations": [
-    "Tune target FPS, queue size, drop policy, or fallback policy for affected agents."
+    "Tune target FPS, queue size, drop policy, or fallback policy for affected agents.",
+    "Lower producer rate, tighten stale-frame drop policy, or move lower priority work behind a fallback path before deployment."
   ],
   "thresholds": {
     "drop_rate_review": 0.2,
@@ -50,7 +76,9 @@
     "runtime_reliability": {
       "drop_rate": 0.5833333333333334,
       "fallback_rate": 0.5833333333333334,
-      "deadline_miss_rate": 0.1
+      "deadline_miss_rate": 0.1,
+      "max_total_queue_depth": 6,
+      "scenario_mode": "sustained_high_load"
     }
   },
   "created_at": "2026-05-17T00:00:00Z"