Add SystemInstructionPart model and support in invocations

instrumentation/opentelemetry-instrumentation-genai-anthropic/src/opentelemetry/instrumentation/genai/anthropic/messages_extractors.py

@@ -25,8 +25,9 @@
 from opentelemetry.util.genai.invocation import InferenceInvocation
 from opentelemetry.util.genai.types import (
     InputMessage,
-    MessagePart,
     OutputMessage,
+    SystemInstructionPart,
+    TextPart,
 )
 from opentelemetry.util.types import AttributeValue
 
@@ -125,10 +126,16 @@ def get_input_messages(
 
 def get_system_instruction(
     system: str | Iterable[TextBlockParam] | None,
-) -> list[MessagePart]:
+) -> list[SystemInstructionPart]:
     if system is None:
         return []
-    return convert_content_to_parts(system)
+    if isinstance(system, str):
+        return [TextPart(content=system)] if system else []
+    return [
+        TextPart(content=block["text"])
+        for block in system
+        if block.get("text")
+    ]
 
 
 def get_output_messages_from_message(

instrumentation/opentelemetry-instrumentation-genai-bedrock/src/opentelemetry/instrumentation/genai/bedrock/extractors.py

@@ -4,6 +4,7 @@
 from __future__ import annotations
 
 import json
+from collections.abc import Mapping, Sequence
 from typing import Any, TypeGuard
 from urllib.parse import urlparse
 
@@ -19,6 +20,7 @@
     OutputMessage,
     ReasoningPart,
     Role,
+    SystemInstructionPart,
     TextPart,
     ToolCallRequestPart,
     ToolCallResponsePart,
@@ -235,7 +237,7 @@ def extract_content_block(block: dict[str, Any]) -> MessagePart | None:
         "toolRemoval",
     ):
         if key in block:
-            return GenericPart(type=key, value=None)
+            return GenericPart(type=key)
 
     return None
 
@@ -256,6 +258,30 @@ def _extract_parts(content: Any) -> list[MessagePart]:
     return parts
 
 
+def _extract_system_parts(
+    content: str | Sequence[Mapping[str, Any] | str] | None,
+) -> list[SystemInstructionPart]:
+    if not content:
+        return []
+    if isinstance(content, str):
+        return [TextPart(content=content)] if content else []
+    parts: list[SystemInstructionPart] = []
+    for item in content:
+        if isinstance(item, str):
+            if item:
+                parts.append(TextPart(content=item))
+        elif _is_dict(item):
+            text = item.get("text")
+            if isinstance(text, str) and text:
+                parts.append(TextPart(content=text))
+            else:
+                for key in item:
+                    if key != "text":
+                        parts.append(GenericPart(type=key))
+                        break
+    return parts
+
+
 def _extract_guardrail_id(
     params: dict[str, Any], invocation: InferenceInvocation
 ) -> None:
@@ -324,7 +350,7 @@ def extract_converse_request(
     # system instruction
     raw_system = kwargs.get("system")
     if capture_content and raw_system:
-        system_parts = _extract_parts(raw_system)
+        system_parts = _extract_system_parts(raw_system)
         if system_parts:
             invocation.system_instruction = system_parts
 
@@ -533,7 +559,7 @@ def extract_invoke_model_request(
     # System instruction (e.g. Anthropic / Nova)
     raw_system = body.get("system")
     if raw_system:
-        system_parts = _extract_parts(raw_system)
+        system_parts = _extract_system_parts(raw_system)
         if system_parts:
             invocation.system_instruction = system_parts
 

instrumentation/opentelemetry-instrumentation-genai-bedrock/tests/test_converse.py

@@ -27,6 +27,7 @@
 )
 from opentelemetry.trace import StatusCode
 from opentelemetry.util.genai.handler import TelemetryHandler
+from opentelemetry.util.genai.types import GenericPart, TextPart
 
 
 @pytest.mark.vcr
@@ -596,3 +597,47 @@ def test_extract_converse_request_prompt_variables_no_content(
         == "sgi5gkybzqak"
     )
     assert "gen_ai.prompt.variable.user_name" not in invocation.attributes
+
+
+def test_extract_converse_request_system_instruction(tracer_provider) -> None:
+    handler = TelemetryHandler(tracer_provider=tracer_provider)
+    invocation = handler.inference(provider="aws.bedrock")
+
+    extract_converse_request(
+        {
+            "system": [
+                {"text": "Be concise"},
+                {"text": "Answer politely"},
+            ],
+        },
+        invocation,
+    )
+
+    assert invocation.system_instruction == [
+        TextPart(content="Be concise"),
+        TextPart(content="Answer politely"),
+    ]
+
+
+def test_extract_converse_request_system_instruction_generic(
+    tracer_provider,
+) -> None:
+    handler = TelemetryHandler(tracer_provider=tracer_provider)
+    invocation = handler.inference(provider="aws.bedrock")
+
+    extract_converse_request(
+        {
+            "system": [
+                {"text": "Be concise"},
+                {"guardContent": {"guardrailIdentifier": "gr-123"}},
+                {"cachePoint": {"type": "default"}},
+            ],
+        },
+        invocation,
+    )
+
+    assert invocation.system_instruction == [
+        TextPart(content="Be concise"),
+        GenericPart(type="guardContent"),
+        GenericPart(type="cachePoint"),
+    ]

instrumentation/opentelemetry-instrumentation-genai-langchain/src/opentelemetry/instrumentation/genai/langchain/callback_handler.py

@@ -40,7 +40,7 @@
     prepare_tool_definitions,
     resolve_response_model_and_id,
     response_fields_from_generation,
-    split_system_and_input_messages,
+    to_input_messages,
 )
 from opentelemetry.util.genai.handler import TelemetryHandler
 from opentelemetry.util.genai.invocation import (
@@ -327,24 +327,21 @@ def on_chat_model_start(
             if "ls_max_tokens" in metadata:
                 max_tokens = metadata.get("ls_max_tokens")
 
-        # Flatten ``list[list[BaseMessage]]`` (one inner list per generation
-        # request) before splitting into system / input.
+        # ``messages`` from on_chat_model_start is ``list[list[BaseMessage]]``
+        # (one inner list per generation request). Flatten and let
+        # :func:`to_input_messages` produce spec-conformant ``InputMessage`` s
+        # with proper roles, tool-call requests, tool results, and reasoning.
         flattened: list[BaseMessage] = [msg for sub in messages for msg in sub]
-        system_instruction: list[MessagePart] = []
         input_messages: list[InputMessage] = []
         if self._telemetry_handler.should_capture_content():
-            system_instruction, input_messages = (
-                split_system_and_input_messages(flattened)
-            )
+            input_messages = to_input_messages(flattened)
 
         llm_invocation = self._telemetry_handler.inference(
             provider,
             request_model=request_model,
         )
         llm_invocation.conversation_id = _conversation_id(metadata)
         llm_invocation.input_messages = input_messages
-        if system_instruction:
-            llm_invocation.system_instruction = system_instruction
         llm_invocation.top_p = top_p
         llm_invocation.frequency_penalty = frequency_penalty
         llm_invocation.presence_penalty = presence_penalty

instrumentation/opentelemetry-instrumentation-genai-langchain/src/opentelemetry/instrumentation/genai/langchain/utils.py

@@ -367,41 +367,6 @@ def to_input_messages(
     return result
 
 
-def split_system_and_input_messages(
-    messages: Iterable[Any],
-) -> tuple[list[MessagePart], list[InputMessage]]:
-    """Split ``messages`` into ``system_instruction`` parts and ``InputMessage`` s.
-
-    Called only when content capture is enabled
-    (``TelemetryHandler.should_capture_content()``).
-    """
-    materialized = list(messages)
-    try:
-        normalized: Iterable[BaseMessage] = convert_to_messages(materialized)
-    except Exception:  # pylint: disable=broad-except
-        normalized = [m for m in materialized if isinstance(m, BaseMessage)]
-
-    system_parts: list[MessagePart] = []
-    input_messages: list[InputMessage] = []
-
-    for message in normalized:
-        if isinstance(message, SystemMessage):
-            system_parts.extend(_content_to_parts(message.content))
-        else:
-            parts = _message_parts(message)
-            if not parts and not _has_content(message):
-                continue
-            input_messages.append(
-                InputMessage(
-                    role=_normalize_role(message) or Role.USER.value,
-                    parts=parts,
-                    name=_message_name(message),
-                )
-            )
-
-    return system_parts, input_messages
-
-
 def to_output_messages(
     messages: Iterable[BaseMessage],
     *,

instrumentation/opentelemetry-instrumentation-genai-langchain/tests/conformance/inference.py

@@ -16,15 +16,12 @@
 from opentelemetry.sdk._logs import LoggerProvider
 from opentelemetry.sdk.metrics import MeterProvider
 from opentelemetry.sdk.trace import TracerProvider
-from opentelemetry.test.weaver_live_check import LiveCheckReport
 from opentelemetry.test_util_genai.conformance import (
     ExpectedViolation,
     Scenario,
 )
 from opentelemetry.test_util_genai.instrumentor import instrument
 
-from ._shared import span_attribute_values
-
 
 class InferenceScenario(Scenario):
     expected_spans = {"chat": 1}
@@ -40,20 +37,6 @@ class InferenceScenario(Scenario):
         ),
     )
 
-    def validate(self, report: LiveCheckReport) -> None:
-        super().validate(report)
-        system_instructions = span_attribute_values(
-            report, "gen_ai.system_instructions"
-        )
-        assert len(system_instructions) == 1, (
-            "chat span with a SystemMessage input should set "
-            f"gen_ai.system_instructions once; saw {system_instructions}"
-        )
-        assert "You are a helpful assistant!" in system_instructions[0], (
-            "gen_ai.system_instructions should carry the SystemMessage "
-            f"content; got {system_instructions[0]}"
-        )
-
     def run(
         self,
         *,

instrumentation/opentelemetry-instrumentation-genai-langchain/tests/conformance/inference_streaming.py

@@ -28,8 +28,6 @@
 )
 from opentelemetry.test_util_genai.instrumentor import instrument
 
-from ._shared import span_attribute_values
-
 
 class InferenceStreamingScenario(Scenario):
     expected_spans = {"chat": 1}
@@ -55,22 +53,17 @@ class InferenceStreamingScenario(Scenario):
 
     def validate(self, report: LiveCheckReport) -> None:
         super().validate(report)
-        stream_values = span_attribute_values(report, "gen_ai.request.stream")
+        stream_values = [
+            attr["value"]
+            for entry in report["samples"]
+            if "span" in entry
+            for attr in entry["span"]["attributes"]
+            if attr["name"] == "gen_ai.request.stream"
+        ]
         assert stream_values == [True], (
             "streaming chat should set gen_ai.request.stream=true on the chat "
             f"span; saw {stream_values}"
         )
-        system_instructions = span_attribute_values(
-            report, "gen_ai.system_instructions"
-        )
-        assert len(system_instructions) == 1, (
-            "streaming chat span with a SystemMessage input should set "
-            f"gen_ai.system_instructions once; saw {system_instructions}"
-        )
-        assert "You are a helpful assistant!" in system_instructions[0], (
-            "gen_ai.system_instructions should carry the SystemMessage "
-            f"content; got {system_instructions[0]}"
-        )
 
     def run(
         self,