Python: Input & output attributes in invoke_agent span (#12834)

### Motivation and Context

<!-- Thank you for your contribution to the semantic-kernel repo!
Please help reviewers and future users, providing the following
information:
  1. Why is this change required?
  2. What problem does it solve?
  3. What scenario does it contribute to?
  4. If it fixes an open issue, please link to the issue here.
-->
The OTel workgroup has been thinking about enhancing tracing for agents.
This is a proposal from our tracing team:
open-telemetry/semantic-conventions#2528. This
PR implements part of the proposal so that when the proposal gets
merged, we can quickly follow.

### Description

<!-- Describe your changes, the overall approach, the underlying design.
These notes will help understanding how your code works. Thanks! -->
Two new attributes are added to the `invoke_agent` span, namely the
`gen_ai.agent.invocation_input` and `gen_ai.agent.invocation_output`.

<img width="747" height="583" alt="image"
src="https://github.com/user-attachments/assets/d8b81f68-915b-48ee-8f86-66a0d4f686cc"
/>


### Contribution Checklist

<!-- Before submitting this PR, please make sure: -->

- [x] The code builds clean without any errors or warnings
- [x] The PR follows the [SK Contribution
Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
and the [pre-submission formatting
script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts)
raises no violations
- [x] All unit tests pass, and I have added new tests where possible
- [x] I didn't break anyone 😄

---------

Co-authored-by: Evan Mattson <[email protected]>

Author: TaoChenOSU

python/samples/getting_started_with_agents/multi_agent_orchestration/observability.py

@@ -0,0 +1,94 @@
+# Copyright (c) Microsoft. All rights reserved.
+
+import logging
+import os
+
+from azure.monitor.opentelemetry.exporter import AzureMonitorLogExporter, AzureMonitorTraceExporter
+from dotenv import load_dotenv
+from opentelemetry import trace
+from opentelemetry._logs import set_logger_provider
+from opentelemetry.sdk._logs import LoggerProvider, LoggingHandler
+from opentelemetry.sdk._logs.export import BatchLogRecordProcessor
+from opentelemetry.sdk.resources import Resource
+from opentelemetry.sdk.trace import TracerProvider
+from opentelemetry.sdk.trace.export import BatchSpanProcessor
+from opentelemetry.semconv.resource import ResourceAttributes
+from opentelemetry.trace import set_tracer_provider
+from opentelemetry.trace.span import format_trace_id
+
+load_dotenv()
+
+APPINSIGHTS_CONNECTION_STRING = os.getenv("APPINSIGHTS_CONNECTION_STRING")
+
+resource = Resource.create({ResourceAttributes.SERVICE_NAME: "multi_agent_orchestration_sample"})
+
+
+def set_up_logging():
+    class KernelFilter(logging.Filter):
+        """A filter to not process records from semantic_kernel."""
+
+        # These are the namespaces that we want to exclude from logging for the purposes of this demo.
+        namespaces_to_exclude: list[str] = [
+            "semantic_kernel.functions.kernel_plugin",
+            "semantic_kernel.prompt_template.kernel_prompt_template",
+        ]
+
+        def filter(self, record):
+            return not any([record.name.startswith(namespace) for namespace in self.namespaces_to_exclude])
+
+    exporters = []
+    exporters.append(AzureMonitorLogExporter(connection_string=APPINSIGHTS_CONNECTION_STRING))
+
+    # Create and set a global logger provider for the application.
+    logger_provider = LoggerProvider(resource=resource)
+    # Log processors are initialized with an exporter which is responsible
+    # for sending the telemetry data to a particular backend.
+    for log_exporter in exporters:
+        logger_provider.add_log_record_processor(BatchLogRecordProcessor(log_exporter))
+    # Sets the global default logger provider
+    set_logger_provider(logger_provider)
+
+    # Create a logging handler to write logging records, in OTLP format, to the exporter.
+    handler = LoggingHandler()
+    # Add filters to the handler to only process records from semantic_kernel.
+    handler.addFilter(logging.Filter("semantic_kernel"))
+    handler.addFilter(KernelFilter())
+    # Attach the handler to the root logger. `getLogger()` with no arguments returns the root logger.
+    # Events from all child loggers will be processed by this handler.
+    logger = logging.getLogger()
+    logger.addHandler(handler)
+    # Set the logging level to NOTSET to allow all records to be processed by the handler.
+    logger.setLevel(logging.NOTSET)
+
+
+def set_up_tracing():
+    exporters = []
+    exporters.append(AzureMonitorTraceExporter(connection_string=APPINSIGHTS_CONNECTION_STRING))
+
+    # Initialize a trace provider for the application. This is a factory for creating tracers.
+    tracer_provider = TracerProvider(resource=resource)
+    # Span processors are initialized with an exporter which is responsible
+    # for sending the telemetry data to a particular backend.
+    for exporter in exporters:
+        tracer_provider.add_span_processor(BatchSpanProcessor(exporter))
+    # Sets the global default tracer provider
+    set_tracer_provider(tracer_provider)
+
+
+def enable_observability(func):
+    """A decorator to enable observability for the samples."""
+
+    async def wrapper(*args, **kwargs):
+        if not APPINSIGHTS_CONNECTION_STRING:
+            # If the connection string is not set, skip observability setup.
+            return await func(*args, **kwargs)
+
+        set_up_logging()
+        set_up_tracing()
+
+        tracer = trace.get_tracer(__name__)
+        with tracer.start_as_current_span("main") as current_span:
+            print(f"Trace ID: {format_trace_id(current_span.get_span_context().trace_id)}")
+            return await func(*args, **kwargs)
+
+    return wrapper

python/samples/getting_started_with_agents/multi_agent_orchestration/step4c_handoff_mix_agent_types.py

@@ -5,6 +5,7 @@
 from azure.ai.projects.aio import AIProjectClient
 from azure.identity.aio import DefaultAzureCredential
 
+from samples.getting_started_with_agents.multi_agent_orchestration.observability import enable_observability
 from semantic_kernel.agents import (
     Agent,
     AzureAIAgent,
@@ -180,6 +181,7 @@ def human_response_function() -> ChatMessageContent:
     return ChatMessageContent(role=AuthorRole.USER, content=user_input)
 
 
+@enable_observability
 async def main():
     """Main function to run the agents."""
     # 0. Initialize the Azure AI agent clients

python/semantic_kernel/agents/azure_ai/azure_ai_agent.py

@@ -58,6 +58,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 from semantic_kernel.utils.telemetry.user_agent import APP_INFO, SEMANTIC_KERNEL_USER_AGENT
 
@@ -832,7 +833,7 @@ async def invoke(
                 # Emit tool-related messages only via callback
                 await on_intermediate_message(message)
 
-    @trace_agent_invocation
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,

python/semantic_kernel/agents/bedrock/bedrock_agent.py

@@ -44,6 +44,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 
 logger = logging.getLogger(__name__)
@@ -458,7 +459,7 @@ async def invoke(
             "Failed to get a response from the agent. Please consider increasing the auto invoke attempts."
         )
 
-    @trace_agent_invocation
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,

python/semantic_kernel/agents/chat_completion/chat_completion_agent.py

@@ -34,6 +34,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 
 if TYPE_CHECKING:
@@ -370,7 +371,7 @@ async def invoke(
         ):
             yield AgentResponseItem(message=response, thread=thread)
 
-    @trace_agent_invocation
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,

python/semantic_kernel/agents/copilot_studio/copilot_studio_agent.py

@@ -7,21 +7,10 @@
 from pathlib import Path
 from typing import TYPE_CHECKING, Any, ClassVar, Literal
 
-from microsoft.agents.copilotstudio.client import (
-    AgentType,
-    CopilotClient,
-    PowerPlatformCloud,
-)
+from microsoft.agents.copilotstudio.client import AgentType, CopilotClient, PowerPlatformCloud
 from microsoft.agents.core.models import ActivityTypes
-from msal import (
-    ConfidentialClientApplication,
-    PublicClientApplication,
-)
-from msal_extensions import (
-    FilePersistence,
-    PersistedTokenCache,
-    build_encrypted_persistence,
-)
+from msal import ConfidentialClientApplication, PublicClientApplication
+from msal_extensions import FilePersistence, PersistedTokenCache, build_encrypted_persistence
 from pydantic import ValidationError
 
 from semantic_kernel.agents import Agent
@@ -47,6 +36,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 
 if sys.version_info >= (3, 12):
@@ -509,6 +499,7 @@ async def invoke(
         ):
             yield AgentResponseItem(message=response, thread=thread)
 
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,

python/semantic_kernel/agents/open_ai/openai_assistant_agent.py

@@ -54,6 +54,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 from semantic_kernel.utils.telemetry.user_agent import APP_INFO, prepend_semantic_kernel_to_user_agent
 
@@ -947,7 +948,7 @@ async def invoke(
                 # Emit tool-related messages only via callback
                 await on_intermediate_message(message)
 
-    @trace_agent_invocation
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,

python/semantic_kernel/agents/open_ai/openai_responses_agent.py

@@ -46,6 +46,7 @@
 from semantic_kernel.utils.telemetry.agent_diagnostics.decorators import (
     trace_agent_get_response,
     trace_agent_invocation,
+    trace_agent_streaming_invocation,
 )
 from semantic_kernel.utils.telemetry.user_agent import APP_INFO, prepend_semantic_kernel_to_user_agent
 
@@ -1040,7 +1041,7 @@ async def invoke(
                 # Emit tool-related messages only via callback
                 await on_intermediate_message(message)
 
-    @trace_agent_invocation
+    @trace_agent_streaming_invocation
     @override
     async def invoke_stream(
         self,