nvidia-nat 1.3.0rc5__py3-none-any.whl → 1.3.1__py3-none-any.whl

nat/agent/rewoo_agent/agent.py

@@ -217,7 +217,7 @@ class ReWOOAgentGraph(BaseAgent):
                 if value is not None:
                     if value == placeholder:
                         tool_input[key] = tool_output
-                    elif placeholder in value:
+                    elif isinstance(value, str) and placeholder in value:
                         # If the placeholder is part of the value, replace it with the stringified output
                         tool_input[key] = value.replace(placeholder, str(tool_output))
 

nat/builder/context.py

@@ -19,6 +19,7 @@ from collections.abc import Awaitable
 from collections.abc import Callable
 from contextlib import contextmanager
 from contextvars import ContextVar
+from functools import cached_property
 
 from nat.builder.intermediate_step_manager import IntermediateStepManager
 from nat.builder.user_interaction_manager import UserInteractionManager
@@ -167,7 +168,7 @@ class Context:
         """
         return UserInteractionManager(self._context_state)
 
-    @property
+    @cached_property
     def intermediate_step_manager(self) -> IntermediateStepManager:
         """
         Retrieves the intermediate step manager instance from the current context state.

nat/builder/intermediate_step_manager.py

@@ -91,7 +91,10 @@ class IntermediateStepManager:
             open_step = self._outstanding_start_steps.pop(payload.UUID, None)
 
             if (open_step is None):
-                logger.warning("Step id %s not found in outstanding start steps", payload.UUID)
+                logger.warning(
+                    "Step id %s not found in outstanding start steps. "
+                    "This may occur if the step was started in a different context or already completed.",
+                    payload.UUID)
                 return
 
             parent_step_id = open_step.step_parent_id
@@ -147,7 +150,8 @@ class IntermediateStepManager:
             if (open_step is None):
                 logger.warning(
                     "Created a chunk for step %s, but no matching start step was found. "
-                    "Chunks must be created with the same ID as the start step.",
+                    "Chunks must be created with the same ID as the start step. "
+                    "This may occur if the step was started in a different context.",
                     payload.UUID)
                 return
 

nat/data_models/api_server.py

@@ -608,6 +608,8 @@ class WebSocketUserInteractionResponseMessage(BaseModel):
     type: typing.Literal[WebSocketMessageType.USER_INTERACTION_MESSAGE]
     id: str = "default"
     thread_id: str = "default"
+    parent_id: str = "default"
+    conversation_id: str | None = None
     content: UserMessageContent
     user: User = User()
     security: Security = Security()

nat/data_models/top_p_mixin.py

@@ -28,10 +28,13 @@ class TopPMixin(
         field_name="top_p",
         default_if_supported=1.0,
         keys=("model_name", "model", "azure_deployment"),
-        unsupported=(re.compile(r"gpt-?5", re.IGNORECASE), ),
+        unsupported=(
+            re.compile(r"gpt-?5", re.IGNORECASE),
+            re.compile(r"claude.?sonnet.?4.5", re.IGNORECASE),
+        ),
 ):
     """
-    Mixin class for top-p configuration. Unsupported on models like gpt-5.
+    Mixin class for top-p configuration. Unsupported on models like gpt-5 and claude-sonnet-4.5.
 
     Attributes:
         top_p: Top-p for distribution sampling. Defaults to 1.0 when supported on the model.

nat/eval/rag_evaluator/evaluate.py

@@ -116,11 +116,14 @@ class RAGEvaluator:
             """Convert NaN or None to 0.0 for safe arithmetic/serialization."""
             return 0.0 if v is None or (isinstance(v, float) and math.isnan(v)) else v
 
-        # Convert from list of dicts to dict of lists, coercing NaN/None to 0.0
+        # Keep original scores (preserving NaN/None) for output
+        original_scores_dict = {metric: [score.get(metric) for score in scores] for metric in scores[0]}
+
+        # Convert from list of dicts to dict of lists, coercing NaN/None to 0.0 for average calculation
         scores_dict = {metric: [_nan_to_zero(score.get(metric)) for score in scores] for metric in scores[0]}
         first_metric_name = list(scores_dict.keys())[0] if scores_dict else None
 
-        # Compute the average of each metric, guarding against empty lists
+        # Compute the average of each metric using cleaned scores (NaN/None -> 0.0)
         average_scores = {
             metric: (sum(values) / len(values) if values else 0.0)
             for metric, values in scores_dict.items()
@@ -137,11 +140,11 @@ class RAGEvaluator:
         else:
             ids = df["user_input"].tolist()  # Use "user_input" as ID fallback
 
-        # Construct EvalOutputItem list
+        # Construct EvalOutputItem list using original scores (preserving NaN/None)
         eval_output_items = [
             EvalOutputItem(
                 id=ids[i],
-                score=_nan_to_zero(getattr(row, first_metric_name, 0.0) if first_metric_name else 0.0),
+                score=original_scores_dict[first_metric_name][i] if first_metric_name else None,
                 reasoning={
                     key:
                         getattr(row, key, None)  # Use getattr to safely access attributes

nat/front_ends/fastapi/fastapi_front_end_config.py

@@ -211,6 +211,13 @@ class FastApiFrontEndConfig(FrontEndBaseConfig, name="fastapi"):
             "Maximum number of async jobs to run concurrently, this controls the number of dask workers created. "
             "This parameter is only used when scheduler_address is `None` and a Dask local cluster is created."),
         ge=1)
+    dask_workers: typing.Literal["threads", "processes"] = Field(
+        default="processes",
+        description=(
+            "Type of Dask workers to use. Options are 'threads' for Threaded Dask workers or 'processes' for "
+            "Process based Dask workers. This parameter is only used when scheduler_address is `None` and a local Dask "
+            "cluster is created."),
+    )
     dask_log_level: str = Field(
         default="WARNING",
         description="Logging level for Dask.",

nat/front_ends/fastapi/fastapi_front_end_plugin.py

@@ -120,18 +120,24 @@ class FastApiFrontEndPlugin(DaskClientMixin, FrontEndBase[FastApiFrontEndConfig]
 
                     from dask.distributed import LocalCluster
 
-                    self._cluster = LocalCluster(processes=True,
+                    use_threads = self.front_end_config.dask_workers == 'threads'
+
+                    # set n_workers to max_running_async_jobs + 1 to allow for one worker to handle the cleanup task
+                    self._cluster = LocalCluster(processes=not use_threads,
                                                  silence_logs=dask_log_level,
-                                                 n_workers=self.front_end_config.max_running_async_jobs,
-                                                 threads_per_worker=1)
+                                                 protocol="tcp",
+                                                 n_workers=self.front_end_config.max_running_async_jobs + 1)
 
                     self._scheduler_address = self._cluster.scheduler.address
 
-                    with self.blocking_client(self._scheduler_address) as client:
-                        # Client.run submits a function to be run on each worker
-                        client.run(self._setup_worker)
+                    if not use_threads and sys.platform != "win32":
+                        with self.blocking_client(self._scheduler_address) as client:
+                            # Client.run submits a function to be run on each worker
+                            client.run(self._setup_worker)
 
-                    logger.info("Created local Dask cluster with scheduler at %s", self._scheduler_address)
+                    logger.info("Created local Dask cluster with scheduler at %s using %s workers",
+                                self._scheduler_address,
+                                self.front_end_config.dask_workers)
 
                 except ImportError:
                     logger.warning("Dask is not installed, async execution and evaluation will not be available.")

nat/front_ends/fastapi/fastapi_front_end_plugin_worker.py

@@ -544,7 +544,8 @@ class FastApiFrontEndPluginWorker(FastApiFrontEndPluginWorkerBase):
         GenerateStreamResponseType = workflow.streaming_output_schema
         GenerateSingleResponseType = workflow.single_output_schema
 
-        if self._dask_available:
+        # Skip async generation for custom routes (those with function_name)
+        if self._dask_available and not hasattr(endpoint, 'function_name'):
             # Append job_id and expiry_seconds to the input schema, this effectively makes these reserved keywords
             # Consider prefixing these with "nat_" to avoid conflicts
 
@@ -562,6 +563,10 @@ class FastApiFrontEndPluginWorker(FastApiFrontEndPluginWorkerBase):
                                             description="Optional time (in seconds) before the job expires. "
                                             "Clamped between 600 (10 min) and 86400 (24h).")
 
+                def validate_model(self):
+                    # Override to ensure that the parent class validator is not called
+                    return self
+
         # Ensure that the input is in the body. POD types are treated as query parameters
         if (not issubclass(GenerateBodyType, BaseModel)):
             GenerateBodyType = typing.Annotated[GenerateBodyType, Body()]
@@ -760,17 +765,18 @@ class FastApiFrontEndPluginWorker(FastApiFrontEndPluginWorkerBase):
                             return AsyncGenerateResponse(job_id=job.job_id, status=job.status)
 
                     job_id = self._job_store.ensure_job_id(request.job_id)
-                    (_, job) = await self._job_store.submit_job(job_id=job_id,
-                                                                expiry_seconds=request.expiry_seconds,
-                                                                job_fn=run_generation,
-                                                                sync_timeout=request.sync_timeout,
-                                                                job_args=[
-                                                                    self._scheduler_address,
-                                                                    self._db_url,
-                                                                    self._config_file_path,
-                                                                    job_id,
-                                                                    request.model_dump(mode="json")
-                                                                ])
+                    (_, job) = await self._job_store.submit_job(
+                        job_id=job_id,
+                        expiry_seconds=request.expiry_seconds,
+                        job_fn=run_generation,
+                        sync_timeout=request.sync_timeout,
+                        job_args=[
+                            self._scheduler_address,
+                            self._db_url,
+                            self._config_file_path,
+                            job_id,
+                            request.model_dump(mode="json", exclude=["job_id", "sync_timeout", "expiry_seconds"])
+                        ])
 
                     if job is not None:
                         response.status_code = 200
@@ -916,7 +922,7 @@ class FastApiFrontEndPluginWorker(FastApiFrontEndPluginWorkerBase):
                     responses={500: response_500},
                 )
 
-                if self._dask_available:
+                if self._dask_available and not hasattr(endpoint, 'function_name'):
                     app.add_api_route(
                         path=f"{endpoint.path}/async",
                         endpoint=post_async_generation(request_type=AsyncGenerateRequest),
@@ -930,7 +936,7 @@ class FastApiFrontEndPluginWorker(FastApiFrontEndPluginWorkerBase):
             else:
                 raise ValueError(f"Unsupported method {endpoint.method}")
 
-            if self._dask_available:
+            if self._dask_available and not hasattr(endpoint, 'function_name'):
                 app.add_api_route(
                     path=f"{endpoint.path}/async/job/{{job_id}}",
                     endpoint=get_async_job_status,

nat/front_ends/mcp/mcp_front_end_config.py

@@ -37,8 +37,11 @@ class MCPFrontEndConfig(FrontEndBaseConfig, name="mcp"):
     port: int = Field(default=9901, description="Port to bind the server to (default: 9901)", ge=0, le=65535)
     debug: bool = Field(default=False, description="Enable debug mode (default: False)")
     log_level: str = Field(default="INFO", description="Log level for the MCP server (default: INFO)")
-    tool_names: list[str] = Field(default_factory=list,
-                                  description="The list of tools MCP server will expose (default: all tools)")
+    tool_names: list[str] = Field(
+        default_factory=list,
+        description="The list of tools MCP server will expose (default: all tools)."
+        "Tool names can be functions or function groups",
+    )
     transport: Literal["sse", "streamable-http"] = Field(
         default="streamable-http",
         description="Transport type for the MCP server (default: streamable-http, backwards compatible with sse)")

nat/front_ends/mcp/mcp_front_end_plugin_worker.py

@@ -234,6 +234,10 @@ class MCPFrontEndPluginWorker(MCPFrontEndPluginWorkerBase):
             filtered_functions: dict[str, Function] = {}
             for function_name, function in functions.items():
                 if function_name in self.front_end_config.tool_names:
+                    # Treat current tool_names as function names, so check if the function name is in the list
+                    filtered_functions[function_name] = function
+                elif any(function_name.startswith(f"{group_name}.") for group_name in self.front_end_config.tool_names):
+                    # Treat tool_names as function group names, so check if the function name starts with the group name
                     filtered_functions[function_name] = function
                 else:
                     logger.debug("Skipping function %s as it's not in tool_names", function_name)

nat/front_ends/mcp/tool_converter.py

@@ -18,9 +18,12 @@ import logging
 from inspect import Parameter
 from inspect import Signature
 from typing import TYPE_CHECKING
+from typing import Any
 
 from mcp.server.fastmcp import FastMCP
 from pydantic import BaseModel
+from pydantic.fields import FieldInfo
+from pydantic_core import PydanticUndefined
 
 from nat.builder.context import ContextState
 from nat.builder.function import Function
@@ -31,6 +34,41 @@ if TYPE_CHECKING:
 
 logger = logging.getLogger(__name__)
 
+# Sentinel: marks "optional; let Pydantic supply default/factory"
+_USE_PYDANTIC_DEFAULT = object()
+
+
+def is_field_optional(field: FieldInfo) -> tuple[bool, Any]:
+    """Determine if a Pydantic field is optional and extract its default value for MCP signatures.
+
+    For MCP tool signatures, we need to distinguish:
+    - Required fields: marked with Parameter.empty
+    - Optional with concrete default: use that default
+    - Optional with factory: use sentinel so Pydantic can apply the factory later
+
+    Args:
+        field: The Pydantic FieldInfo to check
+
+    Returns:
+        A tuple of (is_optional, default_value):
+        - (False, Parameter.empty) for required fields
+        - (True, actual_default) for optional fields with explicit defaults
+        - (True, _USE_PYDANTIC_DEFAULT) for optional fields with default_factory
+    """
+    if field.is_required():
+        return False, Parameter.empty
+
+    # Field is optional - has either default or factory
+    if field.default is not PydanticUndefined:
+        return True, field.default
+
+    # Factory case: mark optional in signature but don't fabricate a value
+    if field.default_factory is not None:
+        return True, _USE_PYDANTIC_DEFAULT
+
+    # Rare corner case: non-required yet no default surfaced
+    return True, _USE_PYDANTIC_DEFAULT
+
 
 def create_function_wrapper(
     function_name: str,
@@ -76,12 +114,15 @@ def create_function_wrapper(
             # Get the field type and convert to appropriate Python type
             field_type = field.annotation
 
+            # Check if field is optional and get its default value
+            _is_optional, param_default = is_field_optional(field)
+
             # Add the parameter to our list
             parameters.append(
                 Parameter(
                     name=name,
                     kind=Parameter.KEYWORD_ONLY,
-                    default=Parameter.empty if field.is_required else None,
+                    default=param_default,
                     annotation=field_type,
                 ))
 
@@ -140,33 +181,23 @@ def create_function_wrapper(
                         result = await call_with_observability(lambda: function.ainvoke(chat_request, to_type=str))
                 else:
                     # Regular handling
-                    # Handle complex input schema - if we extracted fields from a nested schema,
-                    # we need to reconstruct the input
-                    if len(schema.model_fields) == 1 and len(parameters) > 1:
-                        # Get the field name from the original schema
-                        field_name = next(iter(schema.model_fields.keys()))
-                        field_type = schema.model_fields[field_name].annotation
-
-                        # If it's a pydantic model, we need to create an instance
-                        if field_type and hasattr(field_type, "model_validate"):
-                            # Create the nested object
-                            nested_obj = field_type.model_validate(kwargs)
-                            # Call with the nested object
-                            kwargs = {field_name: nested_obj}
+                    # Strip sentinel values so Pydantic can apply defaults/factories
+                    cleaned_kwargs = {k: v for k, v in kwargs.items() if v is not _USE_PYDANTIC_DEFAULT}
+
+                    # Always validate with the declared schema
+                    # This handles defaults, factories, nested models, validators, etc.
+                    model_input = schema.model_validate(cleaned_kwargs)
 
                     # Call the NAT function with the parameters - special handling for Workflow
                     if is_workflow:
-                        # For workflow with regular input, we'll assume the first parameter is the input
-                        input_value = list(kwargs.values())[0] if kwargs else ""
-
-                        # Workflows have a run method that is an async context manager
-                        # that returns a Runner
-                        async with function.run(input_value) as runner:
+                        # Workflows expect the model instance directly
+                        async with function.run(model_input) as runner:
                             # Get the result from the runner
                             result = await runner.result(to_type=str)
                     else:
-                        # Regular function call
-                        result = await call_with_observability(lambda: function.acall_invoke(**kwargs))
+                        # Regular function call - unpack the validated model
+                        result = await call_with_observability(lambda: function.acall_invoke(**model_input.model_dump())
+                                                               )
 
                 # Report completion
                 if ctx:

nat/profiler/parameter_optimization/parameter_optimizer.py

@@ -117,10 +117,21 @@ def optimize_parameters(
 
     # Save final results (out_dir already created and defined above)
     with (out_dir / "optimized_config.yml").open("w") as fh:
-        yaml.dump(tuned_cfg.model_dump(), fh)
+        yaml.dump(tuned_cfg.model_dump(mode='json'), fh)
     with (out_dir / "trials_dataframe_params.csv").open("w") as fh:
         # Export full trials DataFrame (values, params, timings, etc.).
         df = study.trials_dataframe()
+
+        # Rename values_X columns to actual metric names
+        metric_names = list(metric_cfg.keys())
+        rename_mapping = {}
+        for i, metric_name in enumerate(metric_names):
+            old_col = f"values_{i}"
+            if old_col in df.columns:
+                rename_mapping[old_col] = f"values_{metric_name}"
+        if rename_mapping:
+            df = df.rename(columns=rename_mapping)
+
         # Normalise rep_scores column naming for convenience.
         if "user_attrs_rep_scores" in df.columns and "rep_scores" not in df.columns:
             df = df.rename(columns={"user_attrs_rep_scores": "rep_scores"})

nat/profiler/parameter_optimization/pareto_visualizer.py

@@ -46,9 +46,13 @@ class ParetoVisualizer:
 
         fig, ax = plt.subplots(figsize=figsize)
 
-        # Extract metric values
-        x_vals = trials_df[f"values_{0}"].values
-        y_vals = trials_df[f"values_{1}"].values
+        # Extract metric values - support both old (values_0) and new (values_metricname) formats
+        x_col = f"values_{self.metric_names[0]}" \
+            if f"values_{self.metric_names[0]}" in trials_df.columns else f"values_{0}"
+        y_col = f"values_{self.metric_names[1]}"\
+            if f"values_{self.metric_names[1]}" in trials_df.columns else f"values_{1}"
+        x_vals = trials_df[x_col].values
+        y_vals = trials_df[y_col].values
 
         # Plot all trials
         ax.scatter(x_vals,
@@ -62,8 +66,8 @@ class ParetoVisualizer:
 
         # Plot Pareto optimal trials if provided
         if pareto_trials_df is not None and not pareto_trials_df.empty:
-            pareto_x = pareto_trials_df[f"values_{0}"].values
-            pareto_y = pareto_trials_df[f"values_{1}"].values
+            pareto_x = pareto_trials_df[x_col].values
+            pareto_y = pareto_trials_df[y_col].values
 
             ax.scatter(pareto_x,
                        pareto_y,
@@ -98,8 +102,8 @@ class ParetoVisualizer:
         ax.grid(True, alpha=0.3)
 
         # Add direction annotations
-        x_annotation = (f"Better {self.metric_names[0]} →"
-                        if self.directions[0] == "minimize" else f"← Better {self.metric_names[0]}")
+        x_annotation = (f"Better {self.metric_names[0]} ←"
+                        if self.directions[0] == "minimize" else f"→ Better {self.metric_names[0]}")
         ax.annotate(x_annotation,
                     xy=(0.02, 0.98),
                     xycoords='axes fraction',
@@ -109,8 +113,8 @@ class ParetoVisualizer:
                     style='italic',
                     bbox=dict(boxstyle="round,pad=0.3", facecolor="wheat", alpha=0.7))
 
-        y_annotation = (f"Better {self.metric_names[1]} ↑"
-                        if self.directions[1] == "minimize" else f"Better {self.metric_names[1]} ↓")
+        y_annotation = (f"Better {self.metric_names[1]} ↓"
+                        if self.directions[1] == "minimize" else f"Better {self.metric_names[1]} ↑")
         ax.annotate(y_annotation,
                     xy=(0.02, 0.02),
                     xycoords='axes fraction',
@@ -145,7 +149,10 @@ class ParetoVisualizer:
         # Normalize values for better visualization
         all_values = []
         for i in range(n_metrics):
-            all_values.append(trials_df[f"values_{i}"].values)
+            # Support both old (values_0) and new (values_metricname) formats
+            col_name = f"values_{self.metric_names[i]}"\
+                if f"values_{self.metric_names[i]}" in trials_df.columns else f"values_{i}"
+            all_values.append(trials_df[col_name].values)
 
         # Normalize each metric to [0, 1] for parallel coordinates
         normalized_values = []
@@ -221,23 +228,31 @@ class ParetoVisualizer:
 
                 if i == j:
                     # Diagonal: histograms
-                    values = trials_df[f"values_{i}"].values
+                    # Support both old (values_0) and new (values_metricname) formats
+                    col_name = f"values_{self.metric_names[i]}"\
+                        if f"values_{self.metric_names[i]}" in trials_df.columns else f"values_{i}"
+                    values = trials_df[col_name].values
                     ax.hist(values, bins=20, alpha=0.7, color='lightblue', edgecolor='navy')
                     if pareto_trials_df is not None and not pareto_trials_df.empty:
-                        pareto_values = pareto_trials_df[f"values_{i}"].values
+                        pareto_values = pareto_trials_df[col_name].values
                         ax.hist(pareto_values, bins=20, alpha=0.8, color='red', edgecolor='darkred')
                     ax.set_xlabel(f"{self.metric_names[i]}")
                     ax.set_ylabel("Frequency")
                 else:
                     # Off-diagonal: scatter plots
-                    x_vals = trials_df[f"values_{j}"].values
-                    y_vals = trials_df[f"values_{i}"].values
+                    # Support both old (values_0) and new (values_metricname) formats
+                    x_col = f"values_{self.metric_names[j]}"\
+                        if f"values_{self.metric_names[j]}" in trials_df.columns else f"values_{j}"
+                    y_col = f"values_{self.metric_names[i]}"\
+                        if f"values_{self.metric_names[i]}" in trials_df.columns else f"values_{i}"
+                    x_vals = trials_df[x_col].values
+                    y_vals = trials_df[y_col].values
 
                     ax.scatter(x_vals, y_vals, alpha=0.6, s=30, c='lightblue', edgecolors='navy', linewidths=0.5)
 
                     if pareto_trials_df is not None and not pareto_trials_df.empty:
-                        pareto_x = pareto_trials_df[f"values_{j}"].values
-                        pareto_y = pareto_trials_df[f"values_{i}"].values
+                        pareto_x = pareto_trials_df[x_col].values
+                        pareto_y = pareto_trials_df[y_col].values
                         ax.scatter(pareto_x,
                                    pareto_y,
                                    alpha=0.9,

nat/tool/code_execution/code_sandbox.py

@@ -92,7 +92,9 @@ class Sandbox(abc.ABC):
             raise ValueError(f"Language {language} not supported")
 
         generated_code = generated_code.strip().strip("`")
-        code_to_execute = textwrap.dedent("""
+        # Use json.dumps to properly escape the generated_code instead of repr()
+        escaped_code = json.dumps(generated_code)
+        code_to_execute = textwrap.dedent(f"""
             import traceback
             import json
             import os
@@ -101,11 +103,6 @@ class Sandbox(abc.ABC):
             import io
             warnings.filterwarnings('ignore')
             os.environ['OPENBLAS_NUM_THREADS'] = '16'
-        """).strip()
-
-        # Use json.dumps to properly escape the generated_code instead of repr()
-        escaped_code = json.dumps(generated_code)
-        code_to_execute += textwrap.dedent(f"""
 
             generated_code = {escaped_code}
 

nat/tool/server_tools.py

@@ -32,14 +32,23 @@ class RequestAttributesTool(FunctionBaseConfig, name="current_request_attributes
 @register_function(config_type=RequestAttributesTool)
 async def current_request_attributes(config: RequestAttributesTool, builder: Builder):
 
+    from pydantic import RootModel
+    from pydantic.types import JsonValue
     from starlette.datastructures import Headers
     from starlette.datastructures import QueryParams
 
-    async def _get_request_attributes(unused: str) -> str:
+    class RequestBody(RootModel[JsonValue]):
+        """
+        Data model that accepts a request body of any valid JSON type.
+        """
+        root: JsonValue
+
+    async def _get_request_attributes(request_body: RequestBody) -> str:
 
         from nat.builder.context import Context
         nat_context = Context.get()
 
+        # Access request attributes from context
         method: str | None = nat_context.metadata.method
         url_path: str | None = nat_context.metadata.url_path
         url_scheme: str | None = nat_context.metadata.url_scheme
@@ -51,6 +60,9 @@ async def current_request_attributes(config: RequestAttributesTool, builder: Bui
         cookies: dict[str, str] | None = nat_context.metadata.cookies
         conversation_id: str | None = nat_context.conversation_id
 
+        # Access the request body data - can be any valid JSON type
+        request_body_data: JsonValue = request_body.root
+
         return (f"Method: {method}, "
                 f"URL Path: {url_path}, "
                 f"URL Scheme: {url_scheme}, "
@@ -60,7 +72,8 @@ async def current_request_attributes(config: RequestAttributesTool, builder: Bui
                 f"Client Host: {client_host}, "
                 f"Client Port: {client_port}, "
                 f"Cookies: {cookies}, "
-                f"Conversation Id: {conversation_id}")
+                f"Conversation Id: {conversation_id}, "
+                f"Request Body: {request_body_data}")
 
     yield FunctionInfo.from_fn(_get_request_attributes,
                                description="Returns the acquired user defined request attributes.")

{nvidia_nat-1.3.0rc5.dist-info → nvidia_nat-1.3.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: nvidia-nat
-Version: 1.3.0rc5
+Version: 1.3.1
 Summary: NVIDIA NeMo Agent toolkit
 Author: NVIDIA Corporation
 Maintainer: NVIDIA Corporation
@@ -84,10 +84,13 @@ Requires-Dist: nvidia-nat-s3; extra == "s3"
 Provides-Extra: semantic-kernel
 Requires-Dist: nvidia-nat-semantic-kernel; extra == "semantic-kernel"
 Provides-Extra: telemetry
+Requires-Dist: nvidia-nat-data-flywheel; extra == "telemetry"
 Requires-Dist: nvidia-nat-opentelemetry; extra == "telemetry"
 Requires-Dist: nvidia-nat-phoenix; extra == "telemetry"
-Requires-Dist: nvidia-nat-weave; extra == "telemetry"
 Requires-Dist: nvidia-nat-ragaai; extra == "telemetry"
+Requires-Dist: nvidia-nat-weave; extra == "telemetry"
+Provides-Extra: test
+Requires-Dist: nvidia-nat-test; extra == "test"
 Provides-Extra: weave
 Requires-Dist: nvidia-nat-weave; extra == "weave"
 Provides-Extra: zep-cloud

{nvidia_nat-1.3.0rc5.dist-info → nvidia_nat-1.3.1.dist-info}/RECORD

@@ -14,7 +14,7 @@ nat/agent/react_agent/register.py,sha256=qkPaK6AvXjolL-q_Z3waVobXDz24GMfuqGqCn-2
 nat/agent/reasoning_agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nat/agent/reasoning_agent/reasoning_agent.py,sha256=fFQtzvaBWtmr_B6S9KSkqAfyl1BdcOc9xkhnbO4O8Pk,9603
 nat/agent/rewoo_agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nat/agent/rewoo_agent/agent.py,sha256=XXgVXY9xwkyxnr093KXUtfgyNxAQbyGAecoGqN5mMLY,26199
+nat/agent/rewoo_agent/agent.py,sha256=ZcTHOX2cj5zJBgy6suLf_DMFEyBLhjQklPrJuMCjAtU,26226
 nat/agent/rewoo_agent/prompt.py,sha256=B0JeL1xDX4VKcShlkkviEcAsOKAwzSlX8NcAQdmUUPw,3645
 nat/agent/rewoo_agent/register.py,sha256=XArlOR37QOBtAvsdKJUjRok5qTmx39S2mJHSteOwU58,9283
 nat/agent/tool_calling_agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -42,7 +42,7 @@ nat/authentication/oauth2/register.py,sha256=7rXhf-ilgSS_bUJsd9pOOCotL1FM8dKUt3k
 nat/builder/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nat/builder/builder.py,sha256=okI3Y101hwF63AwazzxiahQx-W9eFZ_SNdFXzDuoftU,11608
 nat/builder/component_utils.py,sha256=gxDhm4NCLI1GU0XL9gFe_gife0oJLwgk_YuABJneFfs,13838
-nat/builder/context.py,sha256=6NQmHfJS0gY4eLU7Xg84olmrgUdtVJcS3gmxc-OADiw,13093
+nat/builder/context.py,sha256=YmgYHzXggPQCOIBj1Mirr7xKM7NvtVr3XrSfn979fDM,13138
 nat/builder/embedder.py,sha256=NPkOEcxt_-wc53QRijCQQDLretRUYHRYaKoYmarmrBk,965
 nat/builder/eval_builder.py,sha256=I-ScvupmorClYoVBIs_PhSsB7Xf9e2nGWe0rCZp3txo,6857
 nat/builder/evaluator.py,sha256=xWHMND2vcAUkdFP7FU3jnVki1rUHeTa0-9saFh2hWKs,1162
@@ -51,7 +51,7 @@ nat/builder/front_end.py,sha256=FCJ87NSshVVuTg8zZrq3YAr_u0RaYVZVcibnqlRFy-M,2173
 nat/builder/function.py,sha256=eZZWLwhphgQTnPvbga8sGleX7HCP46usZPIegE7zFzs,27725
 nat/builder/function_base.py,sha256=0Eg8RtjWhEU3Yme0CVxcRutobA0Qo8-YHZLI6L2qAgM,13116
 nat/builder/function_info.py,sha256=7Rmrn-gOFrT2TIJklJwA_O-ycx_oimwZ0-qMYpbuZrU,25161
-nat/builder/intermediate_step_manager.py,sha256=iOuMLWTaES0J0XzaLxhTUqFvuoCAChJu3V69T43K0k0,7599
+nat/builder/intermediate_step_manager.py,sha256=wTfCV47IKpYO7NvoHSD4BGwOiuZ8Db9ktzs14T38gBE,7834
 nat/builder/llm.py,sha256=DW-2q64A06VChsXNEL5PfBjH3DcsnTKVoCEWDuP7MF4,951
 nat/builder/retriever.py,sha256=ZyEqc7pFK31t_yr6Jaxa34c-tRas2edKqJZCNiVh9-0,970
 nat/builder/user_interaction_manager.py,sha256=-Z2qbQes7a2cuXgT7KEbWeuok0HcCnRdw9WB8Ghyl9k,3081
@@ -112,7 +112,7 @@ nat/control_flow/router_agent/prompt.py,sha256=fIAiNsAs1zXRAatButR76zSpHJNxSkXXK
 nat/control_flow/router_agent/register.py,sha256=4RGmS9sy-QtIMmvh8mfMcR1VqxFPLpG4RckWCIExh40,4144
 nat/data_models/__init__.py,sha256=Xs1JQ16L9btwreh4pdGKwskffAw1YFO48jKrU4ib_7c,685
 nat/data_models/agent.py,sha256=IwDyb9Zc3R4Zd5rFeqt7q0EQswczAl5focxV9KozIzs,1625
-nat/data_models/api_server.py,sha256=IowzLwxJqnSkUehCbK0WJp98hBZFXUQDd1cq8lr9PVs,30582
+nat/data_models/api_server.py,sha256=oQtSiP7jpkHIZ75g21A_lTiidNsQo54pq3qy2StIJcs,30652
 nat/data_models/authentication.py,sha256=XPu9W8nh4XRSuxPv3HxO-FMQ_JtTEoK6Y02JwnzDwTg,8457
 nat/data_models/common.py,sha256=nXXfGrjpxebzBUa55mLdmzePLt7VFHvTAc6Znj3yEv0,5875
 nat/data_models/component.py,sha256=b_hXOA8Gm5UNvlFkAhsR6kEvf33ST50MKtr5kWf75Ao,1894
@@ -147,7 +147,7 @@ nat/data_models/swe_bench_model.py,sha256=uZs-hLFuT1B5CiPFwFg1PHinDW8PHne8TBzu7t
 nat/data_models/telemetry_exporter.py,sha256=P7kqxIQnFVuvo_UFpH9QSB8fACy_0U2Uzkw_IfWXagE,998
 nat/data_models/temperature_mixin.py,sha256=LlpfWrWtDrPJfSKfNx5E0P3p5SNGZli7ACRRpmO0QqA,1628
 nat/data_models/thinking_mixin.py,sha256=VRDUJZ8XP_Vv0gW2FRZUf8O9-kVgNEdZCEZ8oEmHyMk,3335
-nat/data_models/top_p_mixin.py,sha256=mu0DLnCAiwNzpSFR8FOW4kQBUpodSrvUR4MsLrNtbgA,1599
+nat/data_models/top_p_mixin.py,sha256=07wOZ7iK2FuNvoyS1LYuIEP0z-7qKDCMoeeeKIAm28k,1705
 nat/data_models/ttc_strategy.py,sha256=tAkKWcyEBmBOOYtHMtQTgeCbHxFTk5SEkmFunNVnfyE,1114
 nat/embedder/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nat/embedder/azure_openai_embedder.py,sha256=8OM54DCVCmWuQ8eZ5lXxsKYQktRmHMoIelRHZmKurUk,2381
@@ -170,7 +170,7 @@ nat/eval/evaluator/__init__.py,sha256=GUJrgGtpvyMUCjUBvR3faAdv-tZzbU9W-izgx9aMEQ
 nat/eval/evaluator/base_evaluator.py,sha256=5WaVGhCGzkynCJyQdxRv7CtqLoUpr6B4O8tilP_gb3g,3232
 nat/eval/evaluator/evaluator_model.py,sha256=riGCcDW8YwC3Kd1yoVmbMdJE1Yf2kVmO8uhsGsKKJA4,1878
 nat/eval/rag_evaluator/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nat/eval/rag_evaluator/evaluate.py,sha256=iQ_fUg1PAuIB2gH6Y7Gz9PeztsdhEswUjyUMa5ZXEx4,8399
+nat/eval/rag_evaluator/evaluate.py,sha256=IfCpfCKBTYhReRkPPbOqyr-9H6gsPGaeFWBIcGDUynw,8639
 nat/eval/rag_evaluator/register.py,sha256=AzT5uICDU5dEo7scvStmOWC7ac-S0Tx4UY87idGtXIs,5835
 nat/eval/runners/__init__.py,sha256=GUJrgGtpvyMUCjUBvR3faAdv-tZzbU9W-izgx9aMEQg,680
 nat/eval/runners/config.py,sha256=bRPai_th02OJrFepbbY6w-t7A18TBXozQUnnnH9iWIU,1403
@@ -239,10 +239,10 @@ nat/front_ends/console/register.py,sha256=2Kf6Mthx6jzWzU8YdhYIR1iABmZDvs1UXM_20n
 nat/front_ends/cron/__init__.py,sha256=Xs1JQ16L9btwreh4pdGKwskffAw1YFO48jKrU4ib_7c,685
 nat/front_ends/fastapi/__init__.py,sha256=Xs1JQ16L9btwreh4pdGKwskffAw1YFO48jKrU4ib_7c,685
 nat/front_ends/fastapi/dask_client_mixin.py,sha256=N_tw4yxA7EKIFTKp5_C2ZksIZucWxRYkFjmZszkAkXc,2072
-nat/front_ends/fastapi/fastapi_front_end_config.py,sha256=BcuzrVlA5b7yYyQKNvQgEanDBtKEHdpC8TAd-O7lfF0,11992
+nat/front_ends/fastapi/fastapi_front_end_config.py,sha256=O_iRpGS3Vpht7ZNr1bYpvoldDEb9Z6f7tFPazVP5i9U,12383
 nat/front_ends/fastapi/fastapi_front_end_controller.py,sha256=ei-34KCMpyaeAgeAN4gVvSGFjewjjRhHZPN0FqAfhDY,2548
-nat/front_ends/fastapi/fastapi_front_end_plugin.py,sha256=e33YkMcLzvm4OUG34bhl-WYiBTqkR-_wJYKG4GODkGM,11169
-nat/front_ends/fastapi/fastapi_front_end_plugin_worker.py,sha256=T6uslFdkHl_r0U54_7cRRKLnWYP2tTMcD7snx9Gv1xs,60547
+nat/front_ends/fastapi/fastapi_front_end_plugin.py,sha256=5akdWipe8onOTdSqrbGq9KO71y0_BNQQJ3JAFj6LmFY,11575
+nat/front_ends/fastapi/fastapi_front_end_plugin_worker.py,sha256=QMERZEifVTDMWttSQqPZu3k8EEPJflPrIR6tkHTkZYg,60594
 nat/front_ends/fastapi/intermediate_steps_subscriber.py,sha256=kbyWlBVpyvyQQjeUnFG9nsR4RaqqNkx567ZSVwwl2RU,3104
 nat/front_ends/fastapi/job_store.py,sha256=cWIBnIgRdkGL7qbBunEKzTYzdPp3l3QCDHMP-qTZJpc,22743
 nat/front_ends/fastapi/main.py,sha256=s8gXCy61rJjK1aywMRpgPvzlkMGsCS-kI_0EIy4JjBM,2445
@@ -259,11 +259,11 @@ nat/front_ends/fastapi/html_snippets/__init__.py,sha256=GUJrgGtpvyMUCjUBvR3faAdv
 nat/front_ends/fastapi/html_snippets/auth_code_grant_success.py,sha256=BNpWwzmA58UM0GK4kZXG4PHJy_5K9ihaVHu8SgCs5JA,1131
 nat/front_ends/mcp/__init__.py,sha256=Xs1JQ16L9btwreh4pdGKwskffAw1YFO48jKrU4ib_7c,685
 nat/front_ends/mcp/introspection_token_verifier.py,sha256=s7Q4Q6rWZJ0ZVujSxxpvVI6Bnhkg1LJQ3RLkvhzFIGE,2836
-nat/front_ends/mcp/mcp_front_end_config.py,sha256=m6z5qSz8YGnFnfu8hRID69suvO1YT_L6sxy1Ki64Ufw,4042
+nat/front_ends/mcp/mcp_front_end_config.py,sha256=KP13AKCyqOwXqebMmYrPwx1Ogq5iPo7R9uzLNSIHySY,4089
 nat/front_ends/mcp/mcp_front_end_plugin.py,sha256=4u_kpen_T-_Uh62V5M7dfW9KyzbqXI7tGBG4AxJXWm0,5231
-nat/front_ends/mcp/mcp_front_end_plugin_worker.py,sha256=jMclC0qEd910oTGCqd1kQ8WjP3WPdQKTl854-2bU_KI,10200
+nat/front_ends/mcp/mcp_front_end_plugin_worker.py,sha256=rOF5Z4o-4n3ky57ncNm8XCdBsbRazGeOeiw55nKbZF4,10618
 nat/front_ends/mcp/register.py,sha256=3aJtgG5VaiqujoeU1-Eq7Hl5pWslIlIwGFU2ASLTXgM,1173
-nat/front_ends/mcp/tool_converter.py,sha256=jyH6tFKUDXSfRBKkv8WjvJsQt05zk3FJBTCwnIuUh5M,11547
+nat/front_ends/mcp/tool_converter.py,sha256=q3GgouOAQ8CGeX8nYtpwHUDEm3BWJgstgZHfKSg2m80,12470
 nat/front_ends/simple_base/__init__.py,sha256=Xs1JQ16L9btwreh4pdGKwskffAw1YFO48jKrU4ib_7c,685
 nat/front_ends/simple_base/simple_front_end_plugin_base.py,sha256=py_yA9XAw-yHfK5cQJLM8ElnubEEM2ac8M0bvz-ScWs,1801
 nat/llm/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -367,9 +367,9 @@ nat/profiler/inference_optimization/experimental/prefix_span_analysis.py,sha256=
 nat/profiler/parameter_optimization/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nat/profiler/parameter_optimization/optimizable_utils.py,sha256=93Pl8A14Zq_f3XsxSH-yFnEJ6B7W5hp7doPnPoLlRB4,3714
 nat/profiler/parameter_optimization/optimizer_runtime.py,sha256=rXmCOq81o7ZorQOUYociVjuO3NO9CIjFBbwql2u_4H4,2715
-nat/profiler/parameter_optimization/parameter_optimizer.py,sha256=CxRFSnl0w0JQvijsY-gszUNxvSrihQ0Jl9ySYClsNFI,6544
+nat/profiler/parameter_optimization/parameter_optimizer.py,sha256=61U2jW0md2oqaN8cy7cmF8oMTbHlgyzdlAXhG5DqfI8,6962
 nat/profiler/parameter_optimization/parameter_selection.py,sha256=pfnNQIx1evNICgChsOJXIFQHoL1R_kmh_vNDsVMC9kg,3982
-nat/profiler/parameter_optimization/pareto_visualizer.py,sha256=IU-4Kw3cVKfDBmXyxtzDf5B325hizMmEYjB9_QJGwD0,15903
+nat/profiler/parameter_optimization/pareto_visualizer.py,sha256=QclLZmmsWINIAh4n0XAKmnIZOqGHTMr-iggZS0kxj-Y,17055
 nat/profiler/parameter_optimization/prompt_optimizer.py,sha256=_AmdeB1jRamd93qR5UqRy5LweYR3bjnD7zoLxzXYE0k,17658
 nat/profiler/parameter_optimization/update_helpers.py,sha256=NxWhrGVchbjws85pPd-jS-C14_l70QvVSvEfENndVcY,2339
 nat/registry_handlers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -420,10 +420,10 @@ nat/tool/github_tools.py,sha256=Pn6p6ebLClNUy6MJIWf-Pl5NguMUT-IJ-bhmpJmvBrg,2190
 nat/tool/nvidia_rag.py,sha256=cEHSc3CZwpd71YcOQngya-Ca_B6ZOb87Dmsoza0VhFY,4163
 nat/tool/register.py,sha256=F1mQs9gI3Ee0EESFyBiTFZeqscY173ENQlfKcWZup0c,1208
 nat/tool/retriever.py,sha256=FP5JL1vCQNrqaKz4F1up-osjxEPhxPFOyaScrgByc34,3877
-nat/tool/server_tools.py,sha256=rQLipwRv8lAyU-gohky2JoVDxWQTUTSttNWjhu7lcHU,3194
+nat/tool/server_tools.py,sha256=sxsgaF5ZjKIc3cSLldt1MDhY3kptrDnkP3kVYvVexfY,3679
 nat/tool/code_execution/README.md,sha256=sl3YX4As95HX61XqTXOGnUcHBV1lla-OeuTnLI4qgng,4019
 nat/tool/code_execution/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-nat/tool/code_execution/code_sandbox.py,sha256=AynQfmvT4mtrV5DHV-20HF8mNTDo1NX4Wj2mlpaEIfs,10159
+nat/tool/code_execution/code_sandbox.py,sha256=evu0n9oTc_MrrQuYbo1TP0sjnjLniRgCsOubO5G5WT4,10090
 nat/tool/code_execution/register.py,sha256=zPFzYqaQhH2B3K8iildVYY_7RKgpoRNKdAo00KmBLQI,3322
 nat/tool/code_execution/test_code_execution_sandbox.py,sha256=iyK9awJs6ST8fc_S3lNPye6It0DxSNp1UIrAUKPJ1h4,14776
 nat/tool/code_execution/utils.py,sha256=__W-T1kaphFKYSc2AydQW8lCdvD7zAccarvs7XVFTtI,4194
@@ -470,10 +470,10 @@ nat/utils/reactive/base/observer_base.py,sha256=6BiQfx26EMumotJ3KoVcdmFBYR_fnAss
 nat/utils/reactive/base/subject_base.py,sha256=UQOxlkZTIeeyYmG5qLtDpNf_63Y7p-doEeUA08_R8ME,2521
 nat/utils/settings/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 nat/utils/settings/global_settings.py,sha256=9JaO6pxKT_Pjw6rxJRsRlFCXdVKCl_xUKU2QHZQWWNM,7294
-nvidia_nat-1.3.0rc5.dist-info/licenses/LICENSE-3rd-party.txt,sha256=fOk5jMmCX9YoKWyYzTtfgl-SUy477audFC5hNY4oP7Q,284609
-nvidia_nat-1.3.0rc5.dist-info/licenses/LICENSE.md,sha256=QwcOLU5TJoTeUhuIXzhdCEEDDvorGiC6-3YTOl4TecE,11356
-nvidia_nat-1.3.0rc5.dist-info/METADATA,sha256=nm0UvCzWa259-7OnT21duAI9yl9EqwKs3lXaxup59OA,10180
-nvidia_nat-1.3.0rc5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-nvidia_nat-1.3.0rc5.dist-info/entry_points.txt,sha256=4jCqjyETMpyoWbCBf4GalZU8I_wbstpzwQNezdAVbbo,698
-nvidia_nat-1.3.0rc5.dist-info/top_level.txt,sha256=lgJWLkigiVZuZ_O1nxVnD_ziYBwgpE2OStdaCduMEGc,8
-nvidia_nat-1.3.0rc5.dist-info/RECORD,,
+nvidia_nat-1.3.1.dist-info/licenses/LICENSE-3rd-party.txt,sha256=fOk5jMmCX9YoKWyYzTtfgl-SUy477audFC5hNY4oP7Q,284609
+nvidia_nat-1.3.1.dist-info/licenses/LICENSE.md,sha256=QwcOLU5TJoTeUhuIXzhdCEEDDvorGiC6-3YTOl4TecE,11356
+nvidia_nat-1.3.1.dist-info/METADATA,sha256=ruhwwSGFasQJ8s7Mai4F3ziiq5VQzCf_d9H7evi4wVA,10308
+nvidia_nat-1.3.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+nvidia_nat-1.3.1.dist-info/entry_points.txt,sha256=4jCqjyETMpyoWbCBf4GalZU8I_wbstpzwQNezdAVbbo,698
+nvidia_nat-1.3.1.dist-info/top_level.txt,sha256=lgJWLkigiVZuZ_O1nxVnD_ziYBwgpE2OStdaCduMEGc,8
+nvidia_nat-1.3.1.dist-info/RECORD,,