Updated UI for streaming

config/agents.yaml

@@ -85,13 +85,26 @@ providers:
 
   # --- Local servers ---
   #
+  # Two variants of the same endpoint: one with Qwen3 thinking mode on, one off.
+  # Qwen3 recommended temperatures: 0.6 (thinking), 0.7 (non-thinking).
+  # Assign reasoning-heavy agents to vastblueai_thinking; procedural/creative
+  # agents to vastblueai.
+  #
   vastblueai:
     type: openai_compatible
     base_url: http://10.250.50.54:9292/v1
     api_key: local
     default_model: "qwen3.5-35-a3b"
     extra_body:
-      enable_thinking: false   # Qwen 3 thinking mode — disable so tokens go to the response
+      enable_thinking: false   # Non-thinking mode — recommended temp 0.7
+
+  vastblueai_thinking:
+    type: openai_compatible
+    base_url: http://10.250.50.54:9292/v1
+    api_key: local
+    default_model: "qwen3.5-35-a3b"
+    extra_body:
+      enable_thinking: true    # Qwen3 thinking mode — recommended temp 0.6
 
   lmstudio:
     type: openai_compatible
@@ -134,19 +147,19 @@ agents:
     title: Chief of Staff
     color: cyan
     prompt_file: miranda_chief_of_staff.md
-    provider: none
+    provider: vastblueai_thinking   # Routing and synthesis benefit from deep reasoning
     model: ""
-    temperature: 0.4
-    max_tokens: 16384
+    temperature: 0.6               # Qwen3 thinking-mode recommendation
+    max_tokens: 32768              # Synthesizes the longest outputs; needs headroom
     stateful: true
 
   vera:
     title: Auditor
     color: yellow
     prompt_file: vera_auditor.md
-    provider: none
+    provider: vastblueai_thinking   # Careful auditing benefits from thinking
     model: ""
-    temperature: 0.2
+    temperature: 0.6               # Qwen3 thinking-mode recommendation
     max_tokens: 16384
     stateful: false
 
@@ -154,9 +167,9 @@ agents:
     title: Director of Personnel & Systems
     color: white
     prompt_file: evelyn_director_of_personnel.md
-    provider: none
+    provider: vastblueai            # Procedural tasks; thinking overhead not warranted
     model: ""
-    temperature: 0.3
+    temperature: 0.7               # Qwen3 non-thinking recommendation
     max_tokens: 16384
     stateful: true
 
@@ -164,9 +177,9 @@ agents:
     title: Director of Research
     color: green
     prompt_file: atlas_research_lead.md
-    provider: none
+    provider: vastblueai_thinking   # Research synthesis benefits from extended reasoning
     model: ""
-    temperature: 0.5
+    temperature: 0.6               # Qwen3 thinking-mode recommendation
     max_tokens: 16384
     stateful: true
 
@@ -174,9 +187,9 @@ agents:
     title: Director of Operations
     color: blue
     prompt_file: cole_operations_lead.md
-    provider: none
+    provider: vastblueai            # Operational tasks are structured and procedural
     model: ""
-    temperature: 0.3
+    temperature: 0.7               # Qwen3 non-thinking recommendation
     max_tokens: 16384
     stateful: true
 
@@ -184,9 +197,9 @@ agents:
     title: Director of Analysis
     color: magenta
     prompt_file: clio_analysis_lead.md
-    provider: none
+    provider: vastblueai_thinking   # Structured analysis benefits from thinking
     model: ""
-    temperature: 0.4
+    temperature: 0.6               # Qwen3 thinking-mode recommendation
     max_tokens: 16384
     stateful: true
 
@@ -194,8 +207,8 @@ agents:
     title: Director of Interface & Experience
     color: bright_cyan
     prompt_file: iris_interface_director.md
-    provider: none
+    provider: vastblueai            # Creative/interface work; non-thinking is sufficient
     model: ""
-    temperature: 0.4
+    temperature: 0.7               # Qwen3 non-thinking recommendation
     max_tokens: 16384
     stateful: true