diff --git a/scripts/ops/hermes_triage_monitor.py b/scripts/ops/hermes_triage_monitor.py
index 4f0b3881a..a38258d48 100755
--- a/scripts/ops/hermes_triage_monitor.py
+++ b/scripts/ops/hermes_triage_monitor.py
@@ -1,16 +1,22 @@
 #!/usr/bin/env python3
 """Stream the triage flow as it happens, in the stages of the flow chart.
 
-Run in a second terminal alongside the demo. Each stage below is a subgraph in
-`mermaid/TestAutomation.mmd`, so what prints here can be followed on the chart:
+Run in a second terminal alongside the demo. Every stage name below is a
+subgraph in `mermaid/TestAutomation.mmd`, and the route line quotes that
+chart's own edge labels, so an audience can follow along on the diagram:
 
     Detect and gather -> Hermes analysis -> Ariadne policy gates
                       -> Ariadne response -> Inspectable outputs
 
-Every stage prints the evidence that moved it, because the point of the chart
-is that each step is answerable from data rather than asserted.
+Each stage prints the evidence that moved it, because the claim the chart
+makes is that every step is answerable from data rather than asserted.
 
-Read-only: it polls Kubernetes, Jenkins and Gitea and changes nothing.
+Two honest limits. The chart draws finer nodes inside each subgraph - the
+collector, the response check, the marker check, the authorizer - and this
+reports at subgraph granularity, not per node. And it follows one incident on
+one job at a time, whereas the chart describes the system as a whole.
+
+Read-only: it polls Kubernetes and changes nothing.
 """
 
 from __future__ import annotations
@@ -32,6 +38,7 @@ STAGES = [
     ("evidence", "Detect and gather", "Ariadne collects the bounded evidence bundle"),
     ("hermes", "Hermes analysis", "Hermes returns a classification it cannot act on"),
     ("gates", "Ariadne policy gates", "Ariadne authorizes or refuses, on its own reading"),
+    ("route", "Ariadne response", "Which branch off the route diamond was taken"),
     ("response", "Ariadne response", "Ariadne executes, proposes, or escalates"),
     ("verify", "Ariadne response", "One rebuild decides whether the repair worked"),
     ("outputs", "Inspectable outputs", "Incident closed; artifacts left behind"),
@@ -150,12 +157,16 @@ def main() -> None:
             if "run_triage" in msg or "hermes run" in msg.lower():
                 monitor.mark("hermes", msg[:110])
             if record.get("event") == "hermes_code_flow":
-                monitor.mark("response", f"code proposal: {record.get('status')}")
+                monitor.mark("route", "code_path (optional source proposal) also taken")
+                monitor.mark("response", f"code proposal: {record.get('status')} -> gitea_pr")
 
         if status == "awaiting_rebuild":
             monitor.mark("evidence", "bundle collected; console regions, tests and logs attached")
             monitor.mark("hermes", "diagnosis returned and parsed against the frozen schema")
             monitor.mark("gates", "every gate passed; Ariadne authorized the predefined action")
+            # The chart labels this edge "authorized action"; naming it lets a
+            # viewer point at the branch being taken rather than infer it.
+            monitor.mark("route", "route --|authorized action|--> registry -> operational_result")
             monitor.mark(
                 "response",
                 f"{state.get('repair', 'action')} on {state.get('target', 'target')}",
@@ -168,6 +179,7 @@ def main() -> None:
                 "gates",
                 f"Ariadne refused: {state.get('reason', 'human_required')} — no action taken",
             )
+            monitor.mark("route", "route --|human required|--> human_required -> gitea_issue")
             monitor.mark("response", "escalated to a human; issue filed in the service repository")
         if status == "healthy" and state.get("resolved") and not monitor.summarised:
             monitor.mark("verify", "rebuild finished green")