Align BaggageBuilder and Outputmessages middleware

[Copilot]

Manual testing
Baggage and outputs can now be set automatically using the middleware.

    "name": "Chat gpt-4o-mini",
    "context": {
        "trace_id": "0x38450fdd1383989d9c2abadd54cef526",
        "span_id": "0x167b776d77c63441",
        "trace_state": "[]"
    },
    "kind": "SpanKind.CLIENT",
    "parent_id": "0x46f72b83108d8113",
    "start_time": "2026-03-02T18:02:00.210340Z",
    "end_time": "2026-03-02T18:02:02.550658Z",
    "status": {
        "status_code": "UNSET"
    },
    "attributes": {
        "operation.source": "SDK",
        "gen_ai.conversation.item.link": "http://localhost:56150/_connector",
        "gen_ai.agent.upn": "b7a3a940-b39a-4a82-9629-0048f947f80a",
        "gen_ai.channel.name": "msteams",
        "gen_ai.system": "az.ai.agent365",
        "gen_ai.agent.id": "84d06baf-416c-4a43-a22d-f2ee93ea9e7e",
        "gen_ai.agent.name": "Azure OpenAI Agent",
        "gen_ai.agent.description": "An AI agent powered by Azure OpenAI",
        "gen_ai.agent.applicationid": "4a380e3b-7092-4d73-bb9d-b6a54702684af",
        "gen_ai.conversation.id": "b84d4c7e-9d83-403f-8f66-22644ff5b0bf",
        "tenant.id": "badf1f56-284d-4dc5-ac59-0dd53900e743",
        "gen_ai.input.messages": "Compute 15 % 4",
        "gen_ai.operation.name": "Chat",
        "gen_ai.request.model": "gpt-4o-mini",
        "gen_ai.provider.name": "Azure OpenAI",
        "gen_ai.output.messages": "\"15 % 4 equals 3. \\n\\nThis is because when you divide 15 by 4, the quotient is 3 and the remainder is 3.\"",
        "gen_ai.usage.input_tokens": "33",
        "gen_ai.usage.output_tokens": "34",
        "gen_ai.response.finish_reasons": "[\"stop\"]"
    },
    "events": [],
    "links": [],
    "resource": {
        "attributes": {
            "telemetry.sdk.language": "python",
            "telemetry.sdk.name": "opentelemetry",
            "telemetry.sdk.version": "1.38.0",
            "service.namespace": "AzureOpenAiKairoTesting",
            "service.name": "AzureOpenAiKairoTracing"
        },
        "schema_url": ""
    }
}
{
    "name": "output_messages 84d06baf-416c-4a43-a22d-f2ee93ea9e7e",
    "context": {
        "trace_id": "0x38450fdd1383989d9c2abadd54cef526",
        "span_id": "0x7f45a1dc264cfc23",
        "trace_state": "[]"
    },
    "kind": "SpanKind.CLIENT",
    "parent_id": "0x46f72b83108d8113",
    "start_time": "2026-03-02T18:02:02.550658Z",
    "end_time": "2026-03-02T18:02:02.552928Z",
    "status": {
        "status_code": "UNSET"
    },
    "attributes": {
        "operation.source": "SDK",
        "gen_ai.conversation.item.link": "http://localhost:56150/_connector",
        "gen_ai.system": "az.ai.agent365",
        "gen_ai.operation.name": "output_messages",
        "gen_ai.agent.id": "84d06baf-416c-4a43-a22d-f2ee93ea9e7e",
        "gen_ai.agent.name": "nikhilcPermTest",
        "gen_ai.agent.description": "agenticUser",
        "gen_ai.agent.upn": "b7a3a940-b39a-4a82-9629-0048f947f80a",
        "tenant.id": "badf1f56-284d-4dc5-ac59-0dd53900e743",
        "gen_ai.output.messages": "[\"15 % 4 equals 3. \\n\\nThis is because when you divide 15 by 4, the quotient is 3 and the remainder is 3.\"]",
        "gen_ai.conversation.id": "b84d4c7e-9d83-403f-8f66-22644ff5b0bf",
        "gen_ai.execution.type": "HumanToAgent",
        "gen_ai.channel.name": "msteams",
        "gen_ai.caller.id": "a92962f3-9ed4-4bcd-9ae0-ad0002b6ca76",
        "gen_ai.caller.name": "Alex Wilber"
    },
    "events": [],
    "links": [],
    "resource": {
        "attributes": {
            "telemetry.sdk.language": "python",
            "telemetry.sdk.name": "opentelemetry",
            "telemetry.sdk.version": "1.38.0",
            "service.namespace": "AzureOpenAiKairoTesting",
            "service.name": "AzureOpenAiKairoTracing"
        },
        "schema_url": ""
    }
}
{
    "name": "invoke_agent Azure OpenAI Agent",
    "context": {
        "trace_id": "0x38450fdd1383989d9c2abadd54cef526",
        "span_id": "0x46f72b83108d8113",
        "trace_state": "[]"
    },
    "kind": "SpanKind.CLIENT",
    "parent_id": null,
    "start_time": "2026-03-02T18:01:59.050044Z",
    "end_time": "2026-03-02T18:02:02.555195Z",
    "status": {
        "status_code": "UNSET"
    },
    "attributes": {
        "operation.source": "SDK",
        "gen_ai.conversation.item.link": "http://localhost:56150/_connector",
        "gen_ai.agent.upn": "b7a3a940-b39a-4a82-9629-0048f947f80a",
        "gen_ai.channel.name": "msteams",
        "gen_ai.caller.id": "a92962f3-9ed4-4bcd-9ae0-ad0002b6ca76",
        "gen_ai.caller.name": "Alex Wilber",
        "gen_ai.system": "az.ai.agent365",
        "gen_ai.operation.name": "invoke_agent",
        "gen_ai.agent.id": "84d06baf-416c-4a43-a22d-f2ee93ea9e7e",
        "gen_ai.agent.name": "Azure OpenAI Agent",
        "gen_ai.agent.description": "An AI agent powered by Azure OpenAI",
        "gen_ai.agent.applicationid": "4a380e3b-7092-4d73-bb9d-b6a54702684af",
        "gen_ai.conversation.id": "b84d4c7e-9d83-403f-8f66-22644ff5b0bf",
        "tenant.id": "badf1f56-284d-4dc5-ac59-0dd53900e743",
        "session.id": "b84d4c7e-9d83-403f-8f66-22644ff5b0bf",
        "gen_ai.execution.type": "HumanToAgent",
        "gen_ai.input.messages": "[\"Compute 15 % 4\"]"
    },
    "events": [],
    "links": [],
    "resource": {
        "attributes": {
            "telemetry.sdk.language": "python",
            "telemetry.sdk.name": "opentelemetry",
            "telemetry.sdk.version": "1.38.0",
            "service.namespace": "AzureOpenAiKairoTesting",
            "service.name": "AzureOpenAiKairoTracing"
        },
        "schema_url": ""
    }
}

The logic callback type in BaggageMiddleware and OutputLoggingMiddleware was changed to Callable[[], Awaitable] to match runtime behavior, but this diverges from the upstream Middleware Protocol which defines Callable[[TurnContext], Awaitable]. All reference implementations (e.g. TranscriptLoggerMiddleware) follow the Protocol signature. Additionally, tests had redundant cases that could be consolidated.

Type annotation

• Reverted logic type to Callable[[TurnContext], Awaitable] in both middlewares to match the Middleware Protocol in microsoft-agents-hosting-core

Agentic gating

• _derive_agent_details gates on activity.is_agentic_request() and returns None for non-agentic requests, preventing spans with empty agent_id
• Test recipient uses role="agenticAppInstance" to exercise the agentic code path

Test consolidation (31 → 28 tests, same coverage)

• Removed test_baggage_middleware_calls_logic — already proven by test_baggage_middleware_propagates_baggage
• Merged test_send_handler_creates_output_scope_for_messages and test_send_handler_disposes_scope_on_success (duplicate success-path setup)
• Merged test_configure_returns_instance into test_configure_is_singleton
• Parametrized 4 middleware registration tests → test_configure_registers_expected_middlewares
• Parametrized 2 TypeError tests → test_configure_raises_on_none

Cleanup

• Removed unused logging/logger from baggage_middleware.py

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.

[github-actions]

⚠️ Deprecation Warning: The deny-licenses option is deprecated for possible removal in the next major release. For more information, see issue 997.

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None

[Copilot]

Pull request overview

This PR updates the hosting-layer observability integration to (1) configure middleware against the concrete MiddlewareSet type with required parameters, and (2) align agentic identity extraction with the canonical Activity helper APIs.

Changes:

• Tightens ObservabilityHostingManager.configure() to require a MiddlewareSet adapter + options (raising TypeError on None) and registers hosting middlewares accordingly.
• Switches agentic property access to Activity.get_agentic_instance_id() / Activity.get_agentic_user() (vs direct recipient.* access).
• Adds/updates unit tests for the new hosting middleware and configuration behavior.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
tests/observability/hosting/scope_helpers/test_scope_helper_utils.py	Updates test data to use an agentic role so agentic helpers are exercised.
tests/observability/hosting/middleware/test_output_logging_middleware.py	Adds tests covering output logging middleware send pipeline behavior.
tests/observability/hosting/middleware/test_observability_hosting_manager.py	Adds tests for singleton configure semantics and required params.
tests/observability/hosting/middleware/test_baggage_middleware.py	Adds tests for baggage propagation and async-reply skip behavior.
tests/observability/hosting/middleware/init.py	Test package init added (copyright header only).
libraries/.../scope_helpers/utils.py	Uses Activity helper methods for agentic instance/user IDs in target agent pairs.
libraries/.../middleware/output_logging_middleware.py	Introduces middleware to create OutputScope spans for outgoing messages.
libraries/.../middleware/observability_hosting_manager.py	Introduces singleton manager that registers hosting observability middleware.
libraries/.../middleware/baggage_middleware.py	Introduces middleware that populates OpenTelemetry baggage from TurnContext.
libraries/.../middleware/init.py	Exports the new middleware + manager symbols.
libraries/.../hosting/init.py	Re-exports middleware/manager at the hosting package level.