more changes

arch/arch_config_schema.yaml

@@ -52,6 +52,8 @@ properties:
               - https
           http_host:
             type: string
+          agent_orchestrator:
+            type: boolean
         additionalProperties: false
         required:
           - endpoint
@@ -93,7 +95,6 @@ properties:
       additionalProperties: false
       required:
         - name
-        - model
   overrides:
     type: object
     properties:

arch/envoy.template.yaml

@@ -142,6 +142,19 @@ static_resources:
                             cluster: {{ llm_cluster_name }}
                             timeout: 60s
                       {% endfor %}
+
+                      {% if agent_orchestrator %}
+                        - match:
+                            prefix: "/"
+                            headers:
+                              - name: "x-arch-llm-provider"
+                                string_match:
+                                  exact: {{ agent_orchestrator }}
+                          route:
+                            auto_host_rewrite: true
+                            cluster: {{ agent_orchestrator }}
+                            timeout: 60s
+                      {% endif %}
                 http_filters:
                   - name: envoy.filters.http.compressor
                     typed_config:

arch/tools/cli/config_generator.py

@@ -150,6 +150,12 @@ def validate_and_render_schema():
     if llm_gateway_listener.get("timeout") == None:
         llm_gateway_listener["timeout"] = "10s"
 
+    agent_orchestrator = None
+    for name, endpoint_details in endpoints.items():
+        if endpoint_details.get("agent_orchestrator", False):
+            agent_orchestrator = name
+            break
+
     data = {
         "prompt_gateway_listener": prompt_gateway_listener,
         "llm_gateway_listener": llm_gateway_listener,
@@ -159,6 +165,7 @@ def validate_and_render_schema():
         "arch_llm_providers": config_yaml["llm_providers"],
         "arch_tracing": arch_tracing,
         "local_llms": llms_with_endpoint,
+        "agent_orchestrator": agent_orchestrator,
     }
 
     rendered = template.render(data)