Responses API: Finish wiring up function tool calls

This finishes the plumbing for function tool call and adds a basic
verification test (that passes for me locally against Llama 4 Scout in
vllm).

Signed-off-by: Ben Browning <bbrownin@redhat.com>

docs/_static/llama-stack-spec.html

@@ -6920,16 +6920,55 @@
                     },
                     {
                         "$ref": "#/components/schemas/OpenAIResponseOutputMessageWebSearchToolCall"
+                    },
+                    {
+                        "$ref": "#/components/schemas/OpenAIResponseOutputMessageFunctionToolCall"
                     }
                 ],
                 "discriminator": {
                     "propertyName": "type",
                     "mapping": {
                         "message": "#/components/schemas/OpenAIResponseMessage",
-                        "web_search_call": "#/components/schemas/OpenAIResponseOutputMessageWebSearchToolCall"
+                        "web_search_call": "#/components/schemas/OpenAIResponseOutputMessageWebSearchToolCall",
+                        "function_call": "#/components/schemas/OpenAIResponseOutputMessageFunctionToolCall"
                     }
                 }
             },
+            "OpenAIResponseOutputMessageFunctionToolCall": {
+                "type": "object",
+                "properties": {
+                    "arguments": {
+                        "type": "string"
+                    },
+                    "call_id": {
+                        "type": "string"
+                    },
+                    "name": {
+                        "type": "string"
+                    },
+                    "type": {
+                        "type": "string",
+                        "const": "function_call",
+                        "default": "function_call"
+                    },
+                    "id": {
+                        "type": "string"
+                    },
+                    "status": {
+                        "type": "string"
+                    }
+                },
+                "additionalProperties": false,
+                "required": [
+                    "arguments",
+                    "call_id",
+                    "name",
+                    "type",
+                    "id",
+                    "status"
+                ],
+                "title": "OpenAIResponseOutputMessageFunctionToolCall"
+            },
             "OpenAIResponseObjectStream": {
                 "oneOf": [
                     {

docs/_static/llama-stack-spec.yaml

@@ -4833,11 +4833,40 @@ components:
       oneOf:
         - $ref: '#/components/schemas/OpenAIResponseMessage'
         - $ref: '#/components/schemas/OpenAIResponseOutputMessageWebSearchToolCall'
+        - $ref: '#/components/schemas/OpenAIResponseOutputMessageFunctionToolCall'
       discriminator:
         propertyName: type
         mapping:
           message: '#/components/schemas/OpenAIResponseMessage'
           web_search_call: '#/components/schemas/OpenAIResponseOutputMessageWebSearchToolCall'
+          function_call: '#/components/schemas/OpenAIResponseOutputMessageFunctionToolCall'
+    "OpenAIResponseOutputMessageFunctionToolCall":
+      type: object
+      properties:
+        arguments:
+          type: string
+        call_id:
+          type: string
+        name:
+          type: string
+        type:
+          type: string
+          const: function_call
+          default: function_call
+        id:
+          type: string
+        status:
+          type: string
+      additionalProperties: false
+      required:
+        - arguments
+        - call_id
+        - name
+        - type
+        - id
+        - status
+      title: >-
+        OpenAIResponseOutputMessageFunctionToolCall
     OpenAIResponseObjectStream:
       oneOf:
         - $ref: '#/components/schemas/OpenAIResponseObjectStreamResponseCreated'

llama_stack/apis/agents/openai_responses.py

@@ -77,8 +77,20 @@ class OpenAIResponseOutputMessageWebSearchToolCall(BaseModel):
     type: Literal["web_search_call"] = "web_search_call"
 
 
+@json_schema_type
+class OpenAIResponseOutputMessageFunctionToolCall(BaseModel):
+    arguments: str
+    call_id: str
+    name: str
+    type: Literal["function_call"] = "function_call"
+    id: str
+    status: str
+
+
 OpenAIResponseOutput = Annotated[
-    OpenAIResponseMessage | OpenAIResponseOutputMessageWebSearchToolCall,
+    OpenAIResponseMessage
+    | OpenAIResponseOutputMessageWebSearchToolCall
+    | OpenAIResponseOutputMessageFunctionToolCall,
     Field(discriminator="type"),
 ]
 register_schema(OpenAIResponseOutput, name="OpenAIResponseOutput")

llama_stack/providers/inline/agents/meta_reference/openai_responses.py

@@ -17,6 +17,7 @@ from llama_stack.apis.agents.openai_responses import (
     OpenAIResponseInputMessageContentImage,
     OpenAIResponseInputMessageContentText,
     OpenAIResponseInputTool,
+    OpenAIResponseInputToolFunction,
     OpenAIResponseMessage,
     OpenAIResponseObject,
     OpenAIResponseObjectStream,
@@ -24,6 +25,7 @@ from llama_stack.apis.agents.openai_responses import (
     OpenAIResponseObjectStreamResponseCreated,
     OpenAIResponseOutput,
     OpenAIResponseOutputMessageContentOutputText,
+    OpenAIResponseOutputMessageFunctionToolCall,
     OpenAIResponseOutputMessageWebSearchToolCall,
     OpenAIResponsePreviousResponseWithInputItems,
 )
@@ -221,10 +223,28 @@ class OpenAIResponsesImpl:
             chat_response = OpenAIChatCompletion(**chat_response.model_dump())
 
         output_messages: list[OpenAIResponseOutput] = []
-        if chat_response.choices[0].message.tool_calls:
+        # TODO: should we check more than choices[0] here?
-            output_messages.extend(
+        if chat_response.choices[0].message.tool_calls and tools:
-                await self._execute_tool_and_return_final_output(model, stream, chat_response, messages, temperature)
+            # TODO: Should we support a mix of custom and builtin tools?
-            )
+            #       in other words, should we check for more than tools[0]?
+            if isinstance(tools[0], OpenAIResponseInputToolFunction):
+                choice = chat_response.choices[0]
+                for tool_call in choice.message.tool_calls:
+                    output_messages.append(
+                        OpenAIResponseOutputMessageFunctionToolCall(
+                            arguments=tool_call.function.arguments or "",
+                            call_id=tool_call.id,
+                            name=tool_call.function.name or "",
+                            id=f"fc_{uuid.uuid4()}",
+                            status="completed",
+                        )
+                    )
+            else:
+                output_messages.extend(
+                    await self._execute_tool_and_return_final_output(
+                        model, stream, chat_response, messages, temperature
+                    )
+                )
         else:
             output_messages.extend(await _openai_choices_to_output_messages(chat_response.choices))
         response = OpenAIResponseObject(

tests/verifications/openai_api/fixtures/test_cases/responses.yaml

@@ -31,6 +31,26 @@ test_response_web_search:
         search_context_size: "low"
       output: "128"
 
+test_response_custom_tool:
+  test_name: test_response_custom_tool
+  test_params:
+    case:
+    - case_id: "sf_weather"
+      input: "What's the weather like in San Francisco?"
+      tools:
+      - type: function
+        name: get_weather
+        description: Get current temperature for a given location.
+        parameters:
+          additionalProperties: false
+          properties:
+            location:
+              description: "City and country e.g. Bogot\xE1, Colombia"
+              type: string
+          required:
+          - location
+          type: object
+
 test_response_image:
   test_name: test_response_image
   test_params:

tests/verifications/openai_api/test_responses.py

@@ -124,6 +124,28 @@ def test_response_non_streaming_web_search(request, openai_client, model, provid
     assert case["output"].lower() in response.output_text.lower().strip()
 
 
+@pytest.mark.parametrize(
+    "case",
+    responses_test_cases["test_response_custom_tool"]["test_params"]["case"],
+    ids=case_id_generator,
+)
+def test_response_non_streaming_custom_tool(request, openai_client, model, provider, verification_config, case):
+    test_name_base = get_base_test_name(request)
+    if should_skip_test(verification_config, provider, model, test_name_base):
+        pytest.skip(f"Skipping {test_name_base} for model {model} on provider {provider} based on config.")
+
+    response = openai_client.responses.create(
+        model=model,
+        input=case["input"],
+        tools=case["tools"],
+        stream=False,
+    )
+    assert len(response.output) == 1
+    assert response.output[0].type == "function_call"
+    assert response.output[0].status == "completed"
+    assert response.output[0].name == "get_weather"
+
+
 @pytest.mark.parametrize(
     "case",
     responses_test_cases["test_response_image"]["test_params"]["case"],