Python API reference

This page covers the public names exported from scouter. For end-to-end setup and examples, start with the guides in the rest of the docs.

Package modules

• alert: Alert configuration objects for Slack, OpsGenie, and console dispatch.
• bifrost: Data access helpers for table-oriented reads and writes.
• client: HTTP client helpers for registering profiles and querying results.
• drift: Drift profile creation and drift result types.
• evaluate: Offline and online evaluation helpers, task types, and result handling.
• agent: Provider integrations, prompt types, and LLM-facing helpers.
• logging: Rust-backed logging configuration and logger access.
• mock: Fixtures and mock utilities used in tests and examples.
• observe: Observation helpers for runtime instrumentation.
• profile: Data profiling classes and feature statistics.
• queue: Queue producers and async ingestion helpers.
• trace: Convenience wrappers for trace context utilities.
• tracing: Tracing runtime, exporters, and instrumentors.
• transport: Transport configuration for HTTP, gRPC, Kafka, RabbitMQ, and Redis.
• types: Shared feature, metric, and alert types.
• util: Utility helpers used by client-side integrations.
• service_map: Service map middleware and connection records.

Top-level exports

Drift and profiling

Drifter

Drifter()

Instantiate Rust Drifter class that is used to create monitoring profiles and compute drifts.

SpcDriftConfig

SpcDriftConfig(space: str = '__missing__', name: str = '__missing__', version: str = '0.1.0', sample_size: int = 25, alert_config: SpcAlertConfig = SpcAlertConfig(), config_path: Optional[Path] = None)

Initialize monitor config

SpcDriftProfile

class SpcDriftProfile

See the signature above and the guide docs for usage examples.

SpcFeatureDriftProfile

class SpcFeatureDriftProfile

See the signature above and the guide docs for usage examples.

SpcFeatureDrift

class SpcFeatureDrift

See the signature above and the guide docs for usage examples.

SpcDriftMap

class SpcDriftMap

Drift map of features

PsiDriftConfig

PsiDriftConfig(space: str = '__missing__', name: str = '__missing__', version: str = '0.1.0', alert_config: PsiAlertConfig = PsiAlertConfig(), config_path: Optional[Path] = None, categorical_features: Optional[list[str]] = None, binning_strategy: QuantileBinning | EqualWidthBinning = QuantileBinning(num_bins=10))

Initialize monitor config

PsiDriftProfile

class PsiDriftProfile

See the signature above and the guide docs for usage examples.

PsiDriftMap

class PsiDriftMap

Drift map of features

FeatureMap

class FeatureMap

See the signature above and the guide docs for usage examples.

CustomMetric

CustomMetric(name: str, baseline_value: float, alert_threshold: AlertThreshold, delta: Optional[float] = None)

Initialize a custom metric for alerting.

CustomDriftProfile

CustomDriftProfile(config: CustomMetricDriftConfig, metrics: list[CustomMetric])

Initialize a CustomDriftProfile instance.

CustomMetricDriftConfig

CustomMetricDriftConfig(space: str = '__missing__', name: str = '__missing__', version: str = '0.1.0', sample_size: int = 25, alert_config: CustomMetricAlertConfig = CustomMetricAlertConfig())

Initialize drift config Args: space: Model space name: Model name version: Model version. Defaults to 0.1.0 sample_size: Sample size alert_config: Custom metric alert configuration

QuantileBinning

QuantileBinning(num_bins: int = 10)

Initialize the quantile binning strategy.

EqualWidthBinning

EqualWidthBinning(method: EqualWidthMethods = Doane())

Initialize the equal-width binning configuration.

Manual

Manual(num_bins: int)

Manual equal-width binning strategy.

SquareRoot

SquareRoot()

Use the SquareRoot equal-width method.

Sturges

Sturges()

Use the Sturges equal-width method.

Rice

Rice()

Use the Rice equal-width method.

Doane

Doane()

Use the Doane equal-width method.

Scott

Scott()

Use the Scott equal-width method.

TerrellScott

TerrellScott()

Use the Terrell-Scott equal-width method.

FreedmanDiaconis

FreedmanDiaconis()

Use the Freedman–Diaconis equal-width method.

DataProfiler

DataProfiler()

Instantiate DataProfiler class that is used to profile data

DataProfile

class DataProfile

Data profile of features

Queue and transport

ScouterQueue

class ScouterQueue

Main queue class for Scouter. Publishes drift records to the configured transport

Queue

class Queue

Individual queue associated with a drift profile

ScouterClient

ScouterClient(config: Optional[HttpConfig] = None)

Helper client for interacting with Scouter Server

DatasetClient

DatasetClient(transport: Any, table_config: Optional[TableConfig] = None)

Dataset client for reading and querying datasets.

DatasetProducer

DatasetProducer(table_config: TableConfig, transport: Any, write_config: Optional[WriteConfig] = None)

Real-time streaming producer for the Scouter dataset engine.

TableConfig

TableConfig(model: Type[Any], catalog: str, schema_name: str, table: str, partition_columns: Optional[List[str]] = None)

Configuration for a dataset table, derived from a Pydantic model.

WriteConfig

WriteConfig(batch_size: int = 1000, scheduled_delay_secs: int = 30)

Configuration for dataset write behavior.

HttpConfig

HttpConfig(server_uri: Optional[str] = None, username: Optional[str] = None, password: Optional[str] = None, auth_token: Optional[str] = None)

HTTP configuration to use with the HTTPProducer.

GrpcConfig

GrpcConfig(server_uri: Optional[str] = None, username: Optional[str] = None, password: Optional[str] = None, timeout_secs: Optional[int] = None, connect_timeout_secs: Optional[int] = None, keep_alive_interval_secs: Optional[int] = None, keep_alive_timeout_secs: Optional[int] = None, keep_alive_while_idle: Optional[bool] = None)

gRPC configuration to use with the GrpcProducer.

KafkaConfig

KafkaConfig(username: Optional[str] = None, password: Optional[str] = None, brokers: Optional[str] = None, topic: Optional[str] = None, compression_type: Optional[str] = None, message_timeout_ms: int = 600000, message_max_bytes: int = 2097164, log_level: LogLevel = LogLevel.Info, config: Dict[str, str] = {}, max_retries: int = 3)

Kafka configuration for connecting to and publishing messages to Kafka brokers.

RabbitMQConfig

RabbitMQConfig(host: Optional[str] = None, port: Optional[int] = None, username: Optional[str] = None, password: Optional[str] = None, queue: Optional[str] = None, max_retries: int = 3)

RabbitMQ configuration to use with the RabbitMQProducer.

RedisConfig

RedisConfig(address: Optional[str] = None, chanel: Optional[str] = None)

Redis configuration to use with a Redis producer

Evaluation

AgentEvalConfig

AgentEvalConfig(space: str = '__missing__', name: str = '__missing__', version: str = '0.1.0', sample_ratio: float = 1.0, alert_config: AgentAlertConfig = AgentAlertConfig())

Initialize drift config Args: space: Space to associate with the config name: Name to associate with the config version: Version to associate with the config. Defaults to 0.1.0 sample_ratio: Sample rate percentage for data collection. Must be between 0.0 and 1.0. Defaults to 1.0 (100%). alert_config: Custom metric alert configuration

AgentEvalProfile

AgentEvalProfile(tasks: _TASK_TYPES, config: Optional[AgentEvalConfig] = None, alias: Optional[str] = None)

Profile for LLM evaluation and drift detection.

EvalRecord

EvalRecord(context: Context, id: Optional[str] = None, session_id: Optional[str] = None, trace_id: Optional[str] = None)

LLM record containing context tied to a Large Language Model interaction that is used to evaluate drift in LLM responses.

LLMJudgeTask

LLMJudgeTask(id: str, prompt: Prompt[Any], expected_value: Any, context_path: Optional[str], operator: ComparisonOperator, description: Optional[str] = None, depends_on: Optional[List[str]] = None, max_retries: Optional[int] = None, condition: bool = False)

LLM-powered evaluation task for complex assessments.

AssertionTask

AssertionTask(id: str, expected_value: Any, operator: ComparisonOperator, context_path: Optional[str] = None, item_context_path: Optional[str] = None, description: Optional[str] = None, depends_on: Optional[Sequence[str]] = None, condition: bool = False)

Assertion-based evaluation task for LLM monitoring.

ComparisonOperator

class ComparisonOperator

Comparison operators for assertion-based evaluations.

EvalResults

class EvalResults

Defines the results of an LLM eval metric

TraceAssertion

class TraceAssertion

Assertion target for trace and span properties.

TraceAssertionTask

TraceAssertionTask(id: str, assertion: TraceAssertion, expected_value: Any, operator: ComparisonOperator, description: Optional[str] = None, depends_on: Optional[List[str]] = None, condition: bool = False)

Trace-based evaluation task for behavioral assertions.

SpanStatus

class SpanStatus

Status codes for trace spans.

AggregationType

class AggregationType

Aggregation operations for span attribute values.

SpanFilter

class SpanFilter

Filter for selecting specific spans within a trace.

Shared types and scheduling

Feature

Feature

No stub docstring was found for this export.

Features

Features(features: List[QueueFeature] | Dict[str, Union[int, float, str]])

Initialize a features class

Metric

Metric(name: str, value: float | int)

Initialize metric

Metrics

Metrics(metrics: List[Metric] | Dict[str, Union[int, float]])

Initialize metrics

CommonCrons

class CommonCrons

See the signature above and the guide docs for usage examples.

PsiAlertConfig

PsiAlertConfig(dispatch_config: Optional[SlackDispatchConfig | OpsGenieDispatchConfig] = None, schedule: Optional[str | CommonCrons] = None, features_to_monitor: List[str] = [], threshold: Optional[PsiThresholdType] = PsiChiSquareThreshold())

Initialize alert config

SpcAlertConfig

SpcAlertConfig(rule: Optional[SpcAlertRule] = None, dispatch_config: Optional[SlackDispatchConfig | OpsGenieDispatchConfig] = None, schedule: Optional[str | CommonCrons] = None, features_to_monitor: List[str] = [])

Initialize alert config

CustomMetricAlertConfig

CustomMetricAlertConfig(dispatch_config: Optional[SlackDispatchConfig | OpsGenieDispatchConfig] = None, schedule: Optional[str | CommonCrons] = None)

Initialize alert config

Bifrost

Bifrost(table_config: TableConfig, transport: Any, write_config: Optional[WriteConfig] = None)

Unified read/write client for the Bifrost dataset engine.

Service map

ServiceConnectionRecord

ServiceConnectionRecord

No stub docstring was found for this export.

ServiceMapMiddleware

ServiceMapMiddleware

No stub docstring was found for this export.