feat: Enable configurable model streaming support (#1202)

Original PR: #1161, close #1099 


https://github.com/user-attachments/assets/95847d85-5062-4755-93b8-0bb07fae9602

- Extends Kagent's ADK OpenAI provider to properly handle streaming with
`LmResponse` partial field, aggregating results, and function calling. I
tested that this works with other model providers as well like Gemini.
- Stream partial events from ADK to the event queue for frontend updates
on LLM response chunks, but also propagate `Event.partial` at the A2A
`Message` level so that the task store will filter them out and avoid
persisting partial chunks (only save `TaskStatusUpdateEvent` that
contain complete chunks).
- An **alternative** would be to mark the partial field at the metadata
at the `Task` level in `_get_context_metadata` and reconstruct the event
in the client side, but currently we do it at the Message level since it
makes more sense. There is a discussion from ADK Python on this
regarding their A2A-ADK integration.
- Update the UI to include a toggle to select whether to use model
streaming. Note that model streaming is different from A2A streaming,
the former streams "chunks" of text, the latter streams tasks.
- Added e2e test that uses `stream: true` supported by the new MockLLM

---------

Signed-off-by: Jet Chiang <pokyuen.jetchiang-ext@solo.io>
Co-authored-by: Azeez Syed <syedazeez337@gmail.com>

Author: supreme-gg-gg

go/api/v1alpha2/agent_types.go

@@ -86,9 +86,9 @@ type DeclarativeAgentSpec struct {
 	// +optional
 	ModelConfig string `json:"modelConfig,omitempty"`
 	// Whether to stream the response from the model.
-	// If not specified, the default value is true.
+	// If not specified, the default value is false.
 	// +optional
-	Stream *bool `json:"stream,omitempty"`
+	Stream bool `json:"stream,omitempty"`
 	// +kubebuilder:validation:MaxItems=20
 	Tools []*Tool `json:"tools,omitempty"`
 	// A2AConfig instantiates an A2A server for this agent,

go/api/v1alpha2/zz_generated.deepcopy.go

@@ -9691,7 +9691,7 @@ spec:
                   stream:
                     description: |-
                       Whether to stream the response from the model.
-                      If not specified, the default value is true.
+                      If not specified, the default value is false.
                     type: boolean
                   systemMessage:
                     description: SystemMessage is a string specifying the system message

go/config/crd/bases/kagent.dev_agents.yaml

@@ -253,6 +253,7 @@ type AgentConfig struct {
 	SseTools     []SseMcpServerConfig  `json:"sse_tools"`
 	RemoteAgents []RemoteAgentConfig   `json:"remote_agents"`
 	ExecuteCode  bool                  `json:"execute_code,omitempty"`
+	Stream       bool                  `json:"stream"`
 }
 
 func (a *AgentConfig) UnmarshalJSON(data []byte) error {

go/internal/adk/types.go

@@ -531,6 +531,7 @@ func (a *adkApiTranslator) translateInlineAgent(ctx context.Context, agent *v1al
 		Instruction: systemMessage,
 		Model:       model,
 		ExecuteCode: false && ptr.Deref(agent.Spec.Declarative.ExecuteCodeBlocks, false), //ignored due to this issue https://github.com/google/adk-python/issues/3921.
+		Stream:      agent.Spec.Declarative.Stream,
 	}
 
 	for _, tool := range agent.Spec.Declarative.Tools {

go/internal/controller/translator/agent/adk_api_translator.go

@@ -0,0 +1,49 @@
+operation: translateAgent
+targetObject: basic-agent
+namespace: test
+objects:
+  - apiVersion: v1
+    kind: Secret
+    metadata:
+      name: openai-secret
+      namespace: test
+    data:
+      api-key: c2stdGVzdC1hcGkta2V5  # base64 encoded "sk-test-api-key"
+  - apiVersion: kagent.dev/v1alpha2
+    kind: ModelConfig
+    metadata:
+      name: basic-model
+      namespace: test
+    spec:
+      provider: OpenAI
+      model: gpt-4o
+      apiKeySecret: openai-secret
+      apiKeySecretKey: api-key
+      openAI:
+        temperature: "0.7"
+        maxTokens: 1024
+        topP: "0.95"
+        reasoningEffort: "low"
+      defaultHeaders:
+        User-Agent: "kagent/1.0"
+  - apiVersion: kagent.dev/v1alpha2
+    kind: Agent
+    metadata:
+      name: basic-agent
+      namespace: test
+    spec:
+      type: Declarative
+      declarative:
+        description: A basic test agent
+        systemMessage: You are a helpful assistant.
+        modelConfig: basic-model
+        stream: true
+        deployment:
+          resources:
+            requests:
+              cpu: 200m
+              memory: 684Mi
+            limits:
+              cpu: 3000m
+              memory: 2Gi
+        tools: []

go/internal/controller/translator/agent/testdata/inputs/agent_with_streaming.yaml

@@ -34,7 +34,8 @@
       "type": "openai"
     },
     "remote_agents": null,
-    "sse_tools": null
+    "sse_tools": null,
+    "stream": false
   },
   "manifest": [
     {
@@ -63,7 +64,7 @@
       },
       "stringData": {
         "agent-card.json": "{\"name\":\"agent_with_code\",\"description\":\"\",\"url\":\"http://agent-with-code.test:8080\",\"version\":\"\",\"capabilities\":{\"streaming\":true,\"pushNotifications\":false,\"stateTransitionHistory\":true},\"defaultInputModes\":[\"text\"],\"defaultOutputModes\":[\"text\"],\"skills\":[]}",
-        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"headers\":{\"User-Agent\":\"kagent/1.0\"},\"base_url\":\"\",\"max_tokens\":1024,\"reasoning_effort\":\"low\",\"temperature\":0.7,\"top_p\":0.95},\"description\":\"\",\"instruction\":\"You are a helpful assistant.\",\"http_tools\":null,\"sse_tools\":null,\"remote_agents\":null}"
+        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"headers\":{\"User-Agent\":\"kagent/1.0\"},\"base_url\":\"\",\"max_tokens\":1024,\"reasoning_effort\":\"low\",\"temperature\":0.7,\"top_p\":0.95},\"description\":\"\",\"instruction\":\"You are a helpful assistant.\",\"http_tools\":null,\"sse_tools\":null,\"remote_agents\":null,\"stream\":false}"
       }
     },
     {
@@ -132,7 +133,7 @@
         "template": {
           "metadata": {
             "annotations": {
-              "kagent.dev/config-hash": "15800629975074137223"
+              "kagent.dev/config-hash": "10194218485187346598"
             },
             "labels": {
               "app": "kagent",

go/internal/controller/translator/agent/testdata/outputs/agent_with_code.json

@@ -41,7 +41,8 @@
       "type": "openai"
     },
     "remote_agents": null,
-    "sse_tools": null
+    "sse_tools": null,
+    "stream": false
   },
   "manifest": [
     {
@@ -70,7 +71,7 @@
       },
       "stringData": {
         "agent-card.json": "{\"name\":\"agent\",\"description\":\"\",\"url\":\"http://agent.test:8080\",\"version\":\"\",\"capabilities\":{\"streaming\":true,\"pushNotifications\":false,\"stateTransitionHistory\":true},\"defaultInputModes\":[\"text\"],\"defaultOutputModes\":[\"text\"],\"skills\":[]}",
-        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a math toolserver. Focus on solving mathematical problems step by step.\",\"http_tools\":[{\"params\":{\"url\":\"http://localhost:8084/mcp\",\"headers\":{\"MATH\":\"sk-test-api-key\"},\"timeout\":30,\"sse_read_timeout\":300},\"tools\":[\"k8s_get_resources\"]}],\"sse_tools\":null,\"remote_agents\":null}"
+        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a math toolserver. Focus on solving mathematical problems step by step.\",\"http_tools\":[{\"params\":{\"url\":\"http://localhost:8084/mcp\",\"headers\":{\"MATH\":\"sk-test-api-key\"},\"timeout\":30,\"sse_read_timeout\":300},\"tools\":[\"k8s_get_resources\"]}],\"sse_tools\":null,\"remote_agents\":null,\"stream\":false}"
       }
     },
     {
@@ -139,7 +140,7 @@
         "template": {
           "metadata": {
             "annotations": {
-              "kagent.dev/config-hash": "8768892159480624774"
+              "kagent.dev/config-hash": "11246224890029939632"
             },
             "labels": {
               "app": "kagent",

go/internal/controller/translator/agent/testdata/outputs/agent_with_http_toolserver.json

@@ -37,7 +37,8 @@
       "type": "openai"
     },
     "remote_agents": null,
-    "sse_tools": null
+    "sse_tools": null,
+    "stream": false
   },
   "manifest": [
     {
@@ -66,7 +67,7 @@
       },
       "stringData": {
         "agent-card.json": "{\"name\":\"agent\",\"description\":\"\",\"url\":\"http://agent.test:8080\",\"version\":\"\",\"capabilities\":{\"streaming\":true,\"pushNotifications\":false,\"stateTransitionHistory\":true},\"defaultInputModes\":[\"text\"],\"defaultOutputModes\":[\"text\"],\"skills\":[]}",
-        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a math toolserver. Focus on solving mathematical problems step by step.\",\"http_tools\":[{\"params\":{\"url\":\"http://toolserver.test:8084/mcp\",\"headers\":{}},\"tools\":[\"k8s_get_resources\"]}],\"sse_tools\":null,\"remote_agents\":null}"
+        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a math toolserver. Focus on solving mathematical problems step by step.\",\"http_tools\":[{\"params\":{\"url\":\"http://toolserver.test:8084/mcp\",\"headers\":{}},\"tools\":[\"k8s_get_resources\"]}],\"sse_tools\":null,\"remote_agents\":null,\"stream\":false}"
       }
     },
     {
@@ -135,7 +136,7 @@
         "template": {
           "metadata": {
             "annotations": {
-              "kagent.dev/config-hash": "12589128794779205881"
+              "kagent.dev/config-hash": "3664580906535457241"
             },
             "labels": {
               "app": "kagent",

go/internal/controller/translator/agent/testdata/outputs/agent_with_mcp_service.json

@@ -35,7 +35,8 @@
         "url": "http://specialist-agent.test:8080"
       }
     ],
-    "sse_tools": null
+    "sse_tools": null,
+    "stream": false
   },
   "manifest": [
     {
@@ -64,7 +65,7 @@
       },
       "stringData": {
         "agent-card.json": "{\"name\":\"parent_agent\",\"description\":\"\",\"url\":\"http://parent-agent.test:8080\",\"version\":\"\",\"capabilities\":{\"streaming\":true,\"pushNotifications\":false,\"stateTransitionHistory\":true},\"defaultInputModes\":[\"text\"],\"defaultOutputModes\":[\"text\"],\"skills\":[]}",
-        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a coordinating agent that can delegate tasks to specialists.\",\"http_tools\":null,\"sse_tools\":null,\"remote_agents\":[{\"name\":\"test__NS__specialist_agent\",\"url\":\"http://specialist-agent.test:8080\",\"headers\":{\"FOO\":\"sup3rs3cr3t\"}}]}"
+        "config.json": "{\"model\":{\"type\":\"openai\",\"model\":\"gpt-4o\",\"base_url\":\"\"},\"description\":\"\",\"instruction\":\"You are a coordinating agent that can delegate tasks to specialists.\",\"http_tools\":null,\"sse_tools\":null,\"remote_agents\":[{\"name\":\"test__NS__specialist_agent\",\"url\":\"http://specialist-agent.test:8080\",\"headers\":{\"FOO\":\"sup3rs3cr3t\"}}],\"stream\":false}"
       }
     },
     {
@@ -133,7 +134,7 @@
         "template": {
           "metadata": {
             "annotations": {
-              "kagent.dev/config-hash": "5107186355478406073"
+              "kagent.dev/config-hash": "10891468836850404374"
             },
             "labels": {
               "app": "kagent",