diff --git a/splunklib/ai/engines/langchain.py b/splunklib/ai/engines/langchain.py
index b53d1cd7..5a090e02 100644
--- a/splunklib/ai/engines/langchain.py
+++ b/splunklib/ai/engines/langchain.py
@@ -286,9 +286,13 @@ async def awrap_tool_call(
                     assert resp.artifact is None, "artifact is already populated"
 
                     if resp.name.startswith(AGENT_PREFIX):
-                        resp.artifact = SubagentFailureResult(str(resp.content))  # pyright: ignore[reportUnknownArgumentType]
+                        resp.artifact = SubagentFailureResult(
+                            error_message=str(resp.content)  # pyright: ignore[reportUnknownArgumentType]
+                        )
                     else:
-                        resp.artifact = ToolFailureResult(str(resp.content))  # pyright: ignore[reportUnknownArgumentType]
+                        resp.artifact = ToolFailureResult(
+                            error_message=str(resp.content)  # pyright: ignore[reportUnknownArgumentType]
+                        )
 
                 return resp
 
@@ -862,7 +866,9 @@ async def llm_handler(req: ModelRequest) -> ModelResponse:
                     case LC_StructuredOutputValidationError():
                         raise StructuredOutputGenerationException(
                             message=msg,
-                            error=StructuredOutputValidationError(str(e.source)),
+                            error=StructuredOutputValidationError(
+                                validation_error=str(e.source)
+                            ),
                         )
                     case LC_StructuredOutputError():
                         # Langchain only returns the above handled exceptions, LC_StructuredOutputError
@@ -1012,7 +1018,7 @@ async def _sdk_handler(request: ToolRequest) -> ToolResponse:
         assert isinstance(sdk_result, ToolMessage), (
             "Expected tool response from tool middleware handler"
         )
-        return ToolResponse(sdk_result.result)
+        return ToolResponse(result=sdk_result.result)
 
     return _sdk_handler
 
@@ -1032,7 +1038,7 @@ async def _sdk_handler(
         assert isinstance(sdk_result, SubagentMessage), (
             "Expected subagent response from subagent middleware handler"
         )
-        return SubagentResponse(sdk_result.result)
+        return SubagentResponse(result=sdk_result.result)
 
     return _sdk_handler
 
@@ -1274,10 +1280,10 @@ def _convert_model_result_from_lc(model_response: LC_ModelCallResult) -> ModelRe
 
         tool_strategy_messages = [
             StructuredOutputMessage(
-                m.tool_call_id,
-                m.name.removeprefix(TOOL_STRATEGY_TOOL_PREFIX) if m.name else "",
-                m.status,
-                str(m.content),  # pyright: ignore[reportUnknownArgumentType]
+                call_id=m.tool_call_id,
+                name=m.name.removeprefix(TOOL_STRATEGY_TOOL_PREFIX) if m.name else "",
+                status=m.status,
+                content=str(m.content),  # pyright: ignore[reportUnknownArgumentType]
             )
             for m in model_response.result
             if isinstance(m, LC_ToolMessage)
@@ -1402,7 +1408,9 @@ async def _tool_call(
                 "ToolException from LangChain should not be raised in tool.func"
             )
 
-        artifact = ToolResult(result.content, result.structured_content)
+        artifact = ToolResult(
+            content=result.content, structured_content=result.structured_content
+        )
 
         if result.structured_content:
             # For both local tools and remote tools (Splunk MCP Server App), the primary
@@ -1719,9 +1727,9 @@ def _map_message_from_langchain(message: LC_BaseMessage) -> BaseMessage:
                 ],
                 structured_output_calls=[
                     StructuredOutputCall(
-                        tc["id"] or "",
-                        tc["name"].removeprefix(TOOL_STRATEGY_TOOL_PREFIX),
-                        tc["args"],
+                        id=tc["id"] or "",
+                        name=tc["name"].removeprefix(TOOL_STRATEGY_TOOL_PREFIX),
+                        args=tc["args"],
                     )
                     for tc in message.tool_calls
                     if tc["name"].startswith(TOOL_STRATEGY_TOOL_PREFIX)
diff --git a/splunklib/ai/messages.py b/splunklib/ai/messages.py
index 614d9d04..4f8ff937 100644
--- a/splunklib/ai/messages.py
+++ b/splunklib/ai/messages.py
@@ -21,7 +21,7 @@
 from splunklib.ai.tools import ToolType
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class TextBlock:
     """Plain text content block returned by a model."""
 
@@ -36,7 +36,7 @@ class TextBlock:
     """
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class OpaqueBlock:
     """Content block of an unrecognized or unsupported type.
 
@@ -62,7 +62,7 @@ class OpaqueBlock:
 ContentBlock = TextBlock | OpaqueBlock
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolCall:
     id: str
     name: str
@@ -70,7 +70,7 @@ class ToolCall:
     args: dict[str, Any]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentCall:
     id: str
     name: str
@@ -78,14 +78,14 @@ class SubagentCall:
     thread_id: str | None
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class StructuredOutputCall:
     id: str
     name: str
     args: dict[str, Any]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class BaseMessage:
     role: str = field(init=False)
 
@@ -96,7 +96,7 @@ def __post_init__(self) -> None:
             )
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class HumanMessage(BaseMessage):
     """
     Message originating from a human user.
@@ -110,7 +110,7 @@ class HumanMessage(BaseMessage):
     content: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class AIMessage(BaseMessage):
     """
     Message produced by an LLM.
@@ -141,7 +141,7 @@ class AIMessage(BaseMessage):
     """
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolResult:
     """
     ToolResult represents a result of a successful tool call.
@@ -151,7 +151,7 @@ class ToolResult:
     structured_content: dict[str, Any] | None
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentStructuredResult:
     """
     SubagentStructuredResult represents a result of a successful subagent call.
@@ -161,7 +161,7 @@ class SubagentStructuredResult:
     structured_output: dict[str, Any]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentTextResult:
     """
     SubagentTextResult represents a result of a successful subagent call.
@@ -171,7 +171,7 @@ class SubagentTextResult:
     content: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolFailureResult:
     """
     Represents the result of a failed sub-agent call.
@@ -183,7 +183,7 @@ class ToolFailureResult:
     error_message: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentFailureResult:
     """
     Represents the result of a failed tool call.
@@ -195,7 +195,7 @@ class SubagentFailureResult:
     error_message: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolMessage(BaseMessage):
     """ToolMessage represents a response of a tool call"""
 
@@ -208,7 +208,7 @@ class ToolMessage(BaseMessage):
 
 
 # TODO: do we have a test that uses this?
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SystemMessage(BaseMessage):
     """
     A message used to prime or control agent behavior.
@@ -218,7 +218,7 @@ class SystemMessage(BaseMessage):
     content: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentMessage(BaseMessage):
     """
     SubagentMessage represents a response of an agent invocation
@@ -231,7 +231,7 @@ class SubagentMessage(BaseMessage):
     result: SubagentStructuredResult | SubagentTextResult | SubagentFailureResult
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class StructuredOutputMessage(BaseMessage):
     """
     StructuredMessage represents a response to the StructuredOutputCall.
@@ -254,7 +254,7 @@ class StructuredOutputMessage(BaseMessage):
 # where developers might want to store messages in say KV store.
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class AgentResponse(Generic[OutputT]):
     # in case output_schema is provided, this will hold the parsed structured output
     structured_output: OutputT
diff --git a/splunklib/ai/middleware.py b/splunklib/ai/middleware.py
index 0231dbb6..e47affbf 100644
--- a/splunklib/ai/middleware.py
+++ b/splunklib/ai/middleware.py
@@ -30,7 +30,7 @@
 )
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class AgentState:
     """AgentState is available through certain middlewares and contains information about the current state of an agent execution."""
 
@@ -42,13 +42,13 @@ class AgentState:
     token_count: int
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolRequest:
     call: ToolCall
     state: AgentState
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolResponse:
     result: ToolResult | ToolFailureResult
 
@@ -56,13 +56,13 @@ class ToolResponse:
 ToolMiddlewareHandler = Callable[[ToolRequest], Awaitable[ToolResponse]]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentRequest:
     call: SubagentCall
     state: AgentState
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class SubagentResponse:
     result: SubagentStructuredResult | SubagentTextResult | SubagentFailureResult
 
@@ -73,13 +73,13 @@ class SubagentResponse:
 ]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ModelRequest:
     system_message: str
     state: AgentState
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ModelResponse:
     message: AIMessage
     structured_output: Any | None = None
@@ -94,7 +94,7 @@ def __post_init__(self) -> None:
 ModelMiddlewareHandler = Callable[[ModelRequest], Awaitable[ModelResponse]]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class AgentRequest:
     messages: Sequence[BaseMessage]
 
diff --git a/splunklib/ai/model.py b/splunklib/ai/model.py
index c701f5d0..2c1f59c7 100644
--- a/splunklib/ai/model.py
+++ b/splunklib/ai/model.py
@@ -18,14 +18,14 @@
 import httpx
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class PredefinedModel:
     """Base class for models that are predefined in the SDK"""
 
     model: str
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class OpenAIModel(PredefinedModel):
     """Predefined OpenAI Model"""
 
@@ -53,7 +53,7 @@ class OpenAIModel(PredefinedModel):
     """
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class AnthropicModel(PredefinedModel):
     """Predefined Anthropic Model"""
 
diff --git a/splunklib/ai/structured_output.py b/splunklib/ai/structured_output.py
index 06fc9635..3c31fd49 100644
--- a/splunklib/ai/structured_output.py
+++ b/splunklib/ai/structured_output.py
@@ -17,12 +17,12 @@
 from splunklib.ai.messages import AIMessage
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class StructuredOutputMultipleToolCallsError:
     pass
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class StructuredOutputValidationError:
     validation_error: str
 
diff --git a/splunklib/ai/tool_settings.py b/splunklib/ai/tool_settings.py
index fe5fdfae..22ce0c9e 100644
--- a/splunklib/ai/tool_settings.py
+++ b/splunklib/ai/tool_settings.py
@@ -18,7 +18,7 @@
 from splunklib.ai.tools import ToolMetadata
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolAllowlist:
     """Holds tool names and tags allowed to be used by Agents.
 
@@ -41,17 +41,17 @@ def is_allowed(self, tool: ToolMetadata) -> bool:
         return self.custom_predicate(tool) if self.custom_predicate else False
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class RemoteToolSettings:
     allowlist: ToolAllowlist
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class LocalToolSettings:
     allowlist: ToolAllowlist
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolSettings:
     local: LocalToolSettings | bool
     """Controls local tool loading (via ``bin/tools.py``).
diff --git a/splunklib/ai/tools.py b/splunklib/ai/tools.py
index 5846f08e..5dbbbd55 100644
--- a/splunklib/ai/tools.py
+++ b/splunklib/ai/tools.py
@@ -38,7 +38,7 @@ class ToolException(Exception):
     """Custom exception to indicate tool execution errors."""
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolResult:
     content: str
     structured_content: dict[str, Any] | None
@@ -49,7 +49,7 @@ class ToolType(Enum):
     REMOTE = "remote"
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class ToolMetadata:
     name: str
     description: str
@@ -58,7 +58,7 @@ class ToolMetadata:
     tags: list[str]
 
 
-@dataclass(frozen=True)
+@dataclass(frozen=True, kw_only=True)
 class Tool(ToolMetadata):
     func: Callable[..., Awaitable[ToolResult]]
 
diff --git a/tests/integration/ai/test_agent_mcp_tools.py b/tests/integration/ai/test_agent_mcp_tools.py
index 7bd4518d..3bca0a63 100644
--- a/tests/integration/ai/test_agent_mcp_tools.py
+++ b/tests/integration/ai/test_agent_mcp_tools.py
@@ -771,7 +771,7 @@ class ToolResults(BaseModel):
                 assert len(agent.tools) == 2
 
                 content = "Call tools to populate output."
-                response = await agent.invoke([HumanMessage(content)])
+                response = await agent.invoke([HumanMessage(content=content)])
                 print(response.structured_output)
                 assert response.structured_output.remote_temperature == "31.5C"
                 assert response.structured_output.local_temperature == "22.1C"
diff --git a/tests/integration/ai/test_conversation_store.py b/tests/integration/ai/test_conversation_store.py
index f4616ca5..da468cb0 100644
--- a/tests/integration/ai/test_conversation_store.py
+++ b/tests/integration/ai/test_conversation_store.py
@@ -129,7 +129,7 @@ async def _agent_middleware(
             if not after_first_call:
                 return AgentResponse(
                     messages=[
-                        HumanMessage("My name is Mike"),
+                        HumanMessage(content="My name is Mike"),
                         AIMessage(content="Hi Mike!", calls=[]),
                     ],
                     structured_output=None,
diff --git a/tests/integration/ai/test_middleware.py b/tests/integration/ai/test_middleware.py
index c90c82ba..e9142cd1 100644
--- a/tests/integration/ai/test_middleware.py
+++ b/tests/integration/ai/test_middleware.py
@@ -191,7 +191,9 @@ async def test_middleware(
 
             call = request.call
             assert call.id, "Invalid call id received"
-            return ToolResponse(ToolResult(content="0.5C", structured_content=None))
+            return ToolResponse(
+                result=ToolResult(content="0.5C", structured_content=None)
+            )
 
         async with Agent(
             model=await self.model(),
@@ -475,7 +477,9 @@ async def test_middleware(
 
             call = request.call
             assert call.id, "Invalid call id received"
-            return SubagentResponse(SubagentTextResult(content="Chris-superstar"))
+            return SubagentResponse(
+                result=SubagentTextResult(content="Chris-superstar")
+            )
 
         async with (
             Agent(
diff --git a/tests/integration/ai/test_structured_output.py b/tests/integration/ai/test_structured_output.py
index 242b5403..edac8cc8 100644
--- a/tests/integration/ai/test_structured_output.py
+++ b/tests/integration/ai/test_structured_output.py
@@ -699,7 +699,7 @@ async def _model_middleware(
                 raise StructuredOutputGenerationException(
                     message=resp.message,
                     error=StructuredOutputValidationError(
-                        "Validation error: name must have ALL letters capitalized"
+                        validation_error="Validation error: name must have ALL letters capitalized"
                     ),
                 )
             return resp
@@ -736,7 +736,7 @@ async def _model_middleware(
                 raise StructuredOutputGenerationException(
                     message=resp.message,
                     error=StructuredOutputValidationError(
-                        "Validation error: name must have ALL letters capitalized"
+                        validation_error="Validation error: name must have ALL letters capitalized"
                     ),
                 )
             return resp
diff --git a/tests/unit/ai/engine/test_langchain_backend.py b/tests/unit/ai/engine/test_langchain_backend.py
index 1daa0add..ea0e199c 100644
--- a/tests/unit/ai/engine/test_langchain_backend.py
+++ b/tests/unit/ai/engine/test_langchain_backend.py
@@ -239,7 +239,7 @@ def test_map_message_from_langchain_tool(self) -> None:
             content="result",
             tool_call_id="call-1",
             status="error",
-            artifact=ToolFailureResult("result"),
+            artifact=ToolFailureResult(error_message="result"),
         )
         mapped = lc._map_message_from_langchain(message)
 
@@ -255,7 +255,7 @@ def test_map_message_from_langchain_subagent(self) -> None:
             content="subagent output",
             tool_call_id="call-2",
             status="error",
-            artifact=SubagentFailureResult("subagent output"),
+            artifact=SubagentFailureResult(error_message="subagent output"),
         )
         mapped = lc._map_message_from_langchain(message)
 
@@ -550,7 +550,7 @@ def test_map_message_to_langchain_tool(self) -> None:
             name="lookup",
             call_id="call-1",
             type=ToolType.REMOTE,
-            result=ToolFailureResult("result"),
+            result=ToolFailureResult(error_message="result"),
         )
         mapped = lc._map_message_to_langchain(message)
 
@@ -562,7 +562,9 @@ def test_map_message_to_langchain_tool(self) -> None:
 
     def test_map_message_to_langchain_subagent(self) -> None:
         message = SubagentMessage(
-            name="My Agent", call_id="call-2", result=SubagentFailureResult("ping")
+            name="My Agent",
+            call_id="call-2",
+            result=SubagentFailureResult(error_message="ping"),
         )
         mapped = lc._map_message_to_langchain(message)
 
diff --git a/tests/unit/ai/test_tool_settings.py b/tests/unit/ai/test_tool_settings.py
index e6d5ac7f..0f592448 100644
--- a/tests/unit/ai/test_tool_settings.py
+++ b/tests/unit/ai/test_tool_settings.py
@@ -67,7 +67,7 @@ def test_filtering(
     initial_tools: Sequence[Tool],
     expected_tools: Sequence[Tool],
 ) -> None:
-    filters = ToolAllowlist(allowed_names, allowed_tags)
+    filters = ToolAllowlist(names=allowed_names, tags=allowed_tags)
     filtered_tools = [t for t in initial_tools if filters.is_allowed(t)]
 
     assert filtered_tools == expected_tools