splits mcp code to assistant/mcp extensions and runs open deep research async with early pass at messaging

.devcontainer/devcontainer.json

@@ -65,7 +65,6 @@
         "ms-python.debugpy",
         "ms-vscode.extension-test-runner",
         "ms-python.python",
-        "ms-toolsai.datawrangler",
         "ms-vscode.makefile-tools",
         "ms-vscode.vscode-node-azure-pack",
         "tamasfe.even-better-toml",

assistants/codespace-assistant/.vscode/settings.json

@@ -9,6 +9,14 @@
   "editor.formatOnType": true,
   "editor.formatOnSave": true,
   "files.eol": "\n",
+  "files.exclude": {
+    "**/.git": true,
+    "**/.svn": true,
+    "**/.hg": true,
+    "**/CVS": true,
+    "**/.DS_Store": true,
+    "**/Thumbs.db": true
+  },
   "files.trimTrailingWhitespace": true,
   "[json]": {
     "editor.defaultFormatter": "esbenp.prettier-vscode",
@@ -25,6 +33,9 @@
   "python.analysis.inlayHints.functionReturnTypes": true,
   "python.analysis.typeCheckingMode": "standard",
   "python.defaultInterpreterPath": "${workspaceFolder}/.venv",
+  "python.testing.pytestEnabled": true,
+  "python.testing.cwd": "${workspaceFolder}",
+  "python.testing.pytestArgs": [],
   "[python]": {
     "editor.defaultFormatter": "charliermarsh.ruff",
     "editor.formatOnSave": true,
@@ -38,8 +49,10 @@
   "ruff.nativeServer": "on",
   "search.exclude": {
     "**/.venv": true,
-    "**/.data": true
+    "**/.data": true,
+    "**/__pycache__": true
   },
+
   // For use with optional extension: "streetsidesoftware.code-spell-checker"
   "cSpell.ignorePaths": [
     ".venv",
@@ -69,6 +82,7 @@
     "pydantic",
     "pyproject",
     "pyright",
+    "pytest",
     "semanticworkbench",
     "tiktoken",
     "updown",

assistants/codespace-assistant/assistant/chat.py

@@ -10,6 +10,7 @@
 
 import deepmerge
 from assistant_extensions.attachments import AttachmentsExtension
+from assistant_extensions.mcp import MCPToolsConfigModel
 from content_safety.evaluators import CombinedContentSafetyEvaluator
 from semantic_workbench_api_model.workbench_model import (
     ConversationEvent,
@@ -27,7 +28,6 @@
 )
 
 from .config import AssistantConfigModel
-from .extensions.tools import ToolsConfigModel
 from .response import respond_to_conversation
 
 logger = logging.getLogger(__name__)
@@ -67,7 +67,7 @@ async def content_evaluator_factory(context: ConversationContext) -> ContentSafe
 )
 
 
-async def tools_config_provider(context: AssistantContext) -> ToolsConfigModel:
+async def tools_config_provider(context: AssistantContext) -> MCPToolsConfigModel:
     return (await assistant_config.get(context)).extensions_config.tools
 
 

assistants/codespace-assistant/assistant/config.py

@@ -6,13 +6,13 @@
     OpenAIClientConfigModel,
 )
 from assistant_extensions.attachments import AttachmentsConfigModel
+from assistant_extensions.mcp import MCPToolsConfigModel
 from content_safety.evaluators import CombinedContentSafetyEvaluatorConfig
 from openai_client import AzureOpenAIServiceConfig, OpenAIRequestConfig
 from pydantic import BaseModel, Field
 from semantic_workbench_assistant.config import UISchema
 
 from . import helpers
-from .extensions.tools import ToolsConfigModel
 
 # The semantic workbench app uses react-jsonschema-form for rendering
 # dynamic configuration forms based on the configuration model and UI schema
@@ -31,12 +31,12 @@
 
 class ExtensionsConfigModel(BaseModel):
     tools: Annotated[
-        ToolsConfigModel,
+        MCPToolsConfigModel,
         Field(
             title="Tools Configuration",
             description="Configuration for the tools.",
         ),
-    ] = ToolsConfigModel()
+    ] = MCPToolsConfigModel()
 
     attachments: Annotated[
         AttachmentsConfigModel,

assistants/codespace-assistant/assistant/response/completion_handler.py

@@ -5,6 +5,7 @@
 from typing import List
 
 import deepmerge
+from assistant_extensions.mcp import ExtendedCallToolRequestParams, MCPSession, handle_mcp_tool_call
 from openai.types.chat import (
     ChatCompletion,
     ChatCompletionToolMessageParam,
@@ -14,18 +15,13 @@
 from semantic_workbench_api_model.workbench_model import (
     MessageType,
     NewConversationMessage,
+    UpdateParticipant,
 )
 from semantic_workbench_assistant.assistant_app import ConversationContext
 
-from assistant.extensions.tools.__model import MCPSession
-
-from ..extensions.tools import (
-    ToolCall,
-    handle_tool_call,
-)
 from .models import StepResult
 from .utils import (
-    extract_content_from_tool_calls,
+    extract_content_from_mcp_tool_calls,
     get_response_duration_message,
     get_token_usage_message,
 )
@@ -55,7 +51,7 @@ async def handle_error(error_message: str) -> StepResult:
         step_result.status = "error"
         return step_result
 
-    # Get service and request configuration for generative model
+    # get service and request configuration for generative model
     generative_request_config = request_config
 
     # get the total tokens used for the completion
@@ -67,10 +63,10 @@ async def handle_error(error_message: str) -> StepResult:
         content = completion.choices[0].message.content
 
     # check if the completion has tool calls
-    tool_calls: list[ToolCall] = []
+    tool_calls: list[ExtendedCallToolRequestParams] = []
     if completion.choices[0].message.tool_calls:
-        ai_context, tool_calls = extract_content_from_tool_calls([
-            ToolCall(
+        ai_context, tool_calls = extract_content_from_mcp_tool_calls([
+            ExtendedCallToolRequestParams(
                 id=tool_call.id,
                 name=tool_call.function.name,
                 arguments=json.loads(
@@ -104,7 +100,7 @@ async def handle_error(error_message: str) -> StepResult:
     deepmerge.always_merger.merge(
         step_result.metadata,
         {
-            "tool_calls": [tool_call.to_dict() for tool_call in tool_calls],
+            "tool_calls": [tool_call.model_dump(mode="json") for tool_call in tool_calls],
         },
     )
 
@@ -170,21 +166,32 @@ async def handle_error(error_message: str) -> StepResult:
         tool_call_count = 0
         for tool_call in tool_calls:
             tool_call_count += 1
-            try:
-                tool_call_result = await handle_tool_call(
-                    mcp_sessions,
-                    tool_call,
-                    f"{metadata_key}:request:tool_call_{tool_call_count}",
-                )
-            except Exception as e:
-                logger.exception(f"Error handling tool call: {e}")
-                return await handle_error("An error occurred while handling the tool call.")
+
+            tool_call_status = f"using tool `{tool_call.name}`"
+            async with context.set_status(f"{tool_call_status}..."):
+
+                async def on_logging_message(msg: str) -> None:
+                    await context.update_participant_me(UpdateParticipant(status=f"{tool_call_status}: {msg}"))
+
+                try:
+                    tool_call_result = await handle_mcp_tool_call(
+                        mcp_sessions,
+                        tool_call,
+                        f"{metadata_key}:request:tool_call_{tool_call_count}",
+                        on_logging_message,
+                    )
+                except Exception as e:
+                    logger.exception(f"Error handling tool call: {e}")
+                    return await handle_error("An error occurred while handling the tool call.")
 
             # Update content and metadata with tool call result metadata
             deepmerge.always_merger.merge(step_result.metadata, tool_call_result.metadata)
 
-            content = (
-                tool_call_result.content if len(tool_call_result.content) > 0 else "[tool call returned no content]"
+            # FIXME only supporting 1 content item and it's text for now, should support other content types/quantity
+            # Get the content from the tool call result
+            content = next(
+                (content_item.text for content_item in tool_call_result.content if content_item.type == "text"),
+                "[tool call returned no content]",
             )
 
             # Add the token count for the tool call result to the total token count