fix(maf): prevent metric inflation and context breaks

instrumentation-loongsuite/loongsuite-instrumentation-microsoft-agent-framework/CHANGELOG.md

@@ -12,6 +12,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Avoid enriching non-MAF spans that use overlapping GenAI operation names.
 - Preserve parent-child context for legacy Microsoft Agent Framework streaming
   agent spans.
+- Stop exporting process-cumulative GenAI ObservableGauges and use Microsoft
+  Agent Framework's native metric instruments instead.
+- Keep non-streaming MAF spans current for nested work and scope Robin provider
+  suppression to the active LLM call when that optional integration is present.
 
 ## Version 0.7.0 (2026-07-03)
 

instrumentation-loongsuite/loongsuite-instrumentation-microsoft-agent-framework/README.rst

@@ -15,11 +15,14 @@ through ``opentelemetry-util-genai`` before spans are ended:
   recording. This ensures exporter snapshots include
   ``gen_ai.span.kind``, ``gen_ai.operation.name``, normalized provider names,
   token usage, finish reasons, and streaming TTFT where MAF exposes enough
-  data.
+  data. Non-streaming spans remain current for the wrapped call, and nested
+  Robin provider instrumentation is suppressed when its optional context key
+  is available.
 * ``MAFSemanticProcessor`` remains as a compatibility layer for MAF workflow,
-  MCP, private-prefix attribute normalization, and the in-process ARMS gauges.
-  Successful spans keep the OpenTelemetry default ``UNSET`` status; failed
-  spans keep MAF's ``ERROR`` status.
+  MCP, and private-prefix attribute normalization. Metrics come from MAF's
+  native counter and histogram instruments; this package does not export
+  process-cumulative ObservableGauges. Successful spans keep the OpenTelemetry
+  default ``UNSET`` status; failed spans keep MAF's ``ERROR`` status.
 * ``react_step_patch`` (opt-in via ``ARMS_MAF_REACT_STEP_ENABLED=true``)
   wraps ``FunctionInvocationLayer.get_response`` so that each LLM round-trip
   inside the ReAct loop emits one ``react step`` span via
@@ -56,6 +59,4 @@ Env                                     Default     Description
 ``ARMS_MAF_INSTRUMENTATION_ENABLED``    ``true``    Master switch; ``false`` disables instrumentation.
 ``ARMS_MAF_SENSITIVE_DATA_ENABLED``     ``false``   Capture inputs/outputs (linked to MAF's sensitive-data option).
 ``ARMS_MAF_REACT_STEP_ENABLED``         ``false``   Emit ``react step`` spans for non-streaming ReAct tool loops.
-``ARMS_MAF_METRICS_ENABLED``            ``true``    Aggregate ARMS GenAI gauges in-process.
-``ARMS_MAF_SLOW_THRESHOLD_MS``          ``1000``    Slow-call threshold in ms.
 ======================================  ==========  ==============================================================

instrumentation-loongsuite/loongsuite-instrumentation-microsoft-agent-framework/src/opentelemetry/instrumentation/microsoft_agent_framework/__init__.py

@@ -18,7 +18,8 @@
 with ``force=True`` so a sticky user-disable does not block us), bridges MAF's
 native span helpers through ``opentelemetry-util-genai`` finish helpers, and
 registers :class:`~.span_processor.MAFSemanticProcessor` for workflow/MCP
-normalization plus metrics aggregation.
+normalization. Microsoft Agent Framework's native counter and histogram
+instruments remain the metric source.
 
 The optional ReAct step patch (``ARMS_MAF_REACT_STEP_ENABLED=true``) wraps the
 ``FunctionInvocationLayer.get_response`` ReAct loop with
@@ -42,9 +43,7 @@
 from opentelemetry.trace import get_tracer_provider
 
 from .config import (
-    get_slow_threshold_ms,
     is_instrumentation_enabled,
-    is_metrics_enabled,
     is_react_step_enabled,
     is_sensitive_data_enabled,
 )
@@ -116,20 +115,19 @@ def _instrument(self, **kwargs: Any) -> None:
         #    invocation finish helpers. This keeps MAF's span lifetime and
         #    streaming cleanup behavior, but writes AGENT/LLM/TOOL semantic
         #    attributes before span.end() creates the exporter snapshot.
-        apply_util_genai_bridge()
+        apply_util_genai_bridge(
+            tracer_provider=tracer_provider,
+            meter_provider=meter_provider,
+        )
 
         # 3) Register the semantic SpanProcessor. MAF uses the standard OTel
         #    TracerProvider (it does not have its own multi-processor), so
         #    ``add_span_processor`` is the right hook.
-        processor = MAFSemanticProcessor(
-            meter_provider=meter_provider,
-            slow_threshold_ms=get_slow_threshold_ms(),
-            metrics_enabled=is_metrics_enabled(default=True),
-            capture_sensitive_data=sensitive,
-        )
+        processor = MAFSemanticProcessor(capture_sensitive_data=sensitive)
         try:
             tracer_provider.add_span_processor(processor)
         except Exception as exc:
+            revert_util_genai_bridge()
             logger.warning("add_span_processor failed: %s", exc)
             raise