Crate legalis_llm 

Legalis-LLM: LLM integration layer for Legalis-RS.

This crate provides an abstraction layer for LLM providers, enabling pluggable AI models (OpenAI, Anthropic, Gemini, Local LLMs).

Modules

ab_testing

A/B testing for prompts.

audio_transcription

Audio transcription utilities for court recordings and legal audio.

chaos

Chaos testing utilities for resilience verification.

dashboard

Custom metrics dashboard system.

embedding_utils

Utilities for working with embeddings.

errors

Error injection for testing error handling paths.

helpers

Helper macros and utilities for common function patterns.

human_feedback

Human feedback integration for RLHF (Reinforcement Learning from Human Feedback).

incremental

Incremental compilation support for law compiler.

inference_rules

Common inference rules

integration_utils

Integration utility functions.

legal_document_analysis

Legal document analysis module for multi-modal inputs.

legal_finetuning

Legal domain fine-tuning utilities.

optimization

Prompt optimization suggestions and analysis.

otel

OpenTelemetry integration for distributed tracing.

pdf_parsing

PDF parsing utilities for legal documents.

pipeline

Compilation pipeline with configurable stages.

processors

Pre and post processors for custom compilation logic.

progress

Progress tracking for LLM operations.

rag_v2

RAG 2.0 features for advanced retrieval.

rules

Advanced validation rules beyond simple JSON schema.

stream_utils

Stream combinators for transforming text streams.

video_analysis

Video analysis utilities for evidence review.

Structs

ABTest

A/B testing framework for comparing fine-tuned models.

ABTestAnalysis

A/B test analysis results.

ABTestResult

A/B test result comparison.

AISuggestion

AWQConfig

AWQ (Activation-aware Weight Quantization) configuration.

AccessControlManager

Access control manager

AccessPermission

AccessRole

AccuracyIssue

Accuracy issue found during validation

AccuracyValidation

Legal accuracy validation result

ActiveLearner

Active learning system with human feedback.

ActiveLearningStats

AdversarialDetector

Adversarial detector for prompt injection and attacks

AgentOrchestrator

Multi-agent orchestrator for coordinating legal analysis tasks.

AgentResult

Result from an agent’s analysis or task.

AgentTaskResult

Agent task result with detailed information.

AggregatedMetrics

Aggregated metrics across multiple requests.

Amendment

Amendment to a legal rule

AnalysisReport

Report from statute analysis.

AnonymizationPipeline

Data anonymization pipeline

AnonymizationRule

AnthropicClient

Anthropic Claude client.

AsyncCache

Async-aware cache using tokio primitives for better async performance.

AttackDetectionResult

Adversarial attack detection result

AttentionWeights

AttributionReport

AuditEntry

AuditLogEntry

Audit log entry for LLM operations.

AuditLogger

Audit logger for tracking LLM operations.

AuditTrail

Audit trail for tracking operations

AutoScaler

Auto-scaler for LLM deployments.

AutoScalingConfig

Auto-scaling configuration.

AutonomousOrchestrator

Autonomous legal intelligence orchestrator.

AutonomousStats

AzureOpenAiClient

Azure OpenAI client.

BatchConfig

Configuration for batch processing.

BatchProcessor

Batch processor for LLM requests.

BatchQueue

Batching queue for LLM requests.

BatchResult

Result of batch processing with statistics.

BehaviorPrediction

Behavior prediction result

BehavioralTendencies

Behavioral tendencies

BenchmarkResult

Result of running a benchmark on a single task.

BenchmarkStats

Aggregated benchmark statistics across multiple tasks.

BenchmarkSuite

Benchmark suite for comparing models.

BenchmarkTask

A benchmark task to evaluate models.

BiasDetectionResult

Result of bias detection.

BiasDetector

Bias detector using pattern matching.

BiasInstance

A specific instance of detected bias.

BreachAnalysis

Breach analysis

BudgetLimit

Budget limits for token usage.

BudgetStatus

Current budget status.

BudgetTracker

Token tracker with budget management.

BusinessOperation

Business operation

CacheConfig

Configuration for the response cache.

CacheStats

Statistics for cache performance.

CacheWarmer

Cache warming utility for preloading common prompts.

CachedLawCompiler

Law compiler with caching support.

CachedProvider

A wrapper around an LLM provider that adds caching.

CachedResponse

A cached response with metadata.

CalibrationBin

A bin in the calibration curve.

CalibrationMetrics

Confidence calibration metrics.

CalibrationPoint

Confidence calibration data point.

CancellableStream

A wrapper that adds cancellation support to text streams.

CancellationContext

Context for managing cancellable LLM operations.

CancellationGuard

Guards a stream with multiple cancellation conditions.

CaseAnalysisOutput

Structured case analysis output.

CaseAnalyzer

Case analyzer that produces structured output.

CaseCitation

Legal citation edge between cases

CaseFeatures

Case features for prediction

CaseLawAnalysis

Case law analysis result.

CaseNode

Legal case node in citation network

CaseOutcome

Case outcome prediction result

CaseOutcomePredictor

Case outcome predictor

CaseParties

Parties in a case.

CaseResult

Result for a single case.

CausalEvent

Causal event in a legal scenario

CausalGraph

Causal graph for legal scenarios

CausalLink

Causal link between events

CausationAnalysis

Result of causation analysis

ChainExecutionResult

Result of executing a prompt chain.

ChainNode

A node in a prompt chain representing a single step.

ChainOfLawPrompter

Chain-of-law prompter for step-by-step legal reasoning.

ChainOfLawResult

Complete chain-of-law reasoning result.

ChainOfThought

Chain-of-thought prompting builder.

ChainOfThoughtBuilder

Chain-of-thought prompt builder

ChainOfThoughtLogger

Chain-of-thought logger.

Change

Change in document

ChangeSummary

Change summary

CircuitBreaker

Circuit breaker for LLM providers.

CircuitBreakerConfig

Configuration for circuit breaker.

Citation

CitationCluster

Citation cluster (community)

CitationEvolution

Citation evolution over time

CitationExtractor

Legal citation extractor.

CitationGroundedPrompter

Citation-grounded generation prompter.

CitationGroundedResult

Citation-grounded generation result.

CitationNetwork

Legal citation network graph

CitationSupport

Citation support for a specific claim.

ClauseClassification

Clause classification result

ClauseClassifier

Clause classifier

ClauseRanker

Clause suggestion ranker.

ClauseTemplate

ClusterStats

Statistics for the inference cluster.

CohereClient

Cohere client for natural language understanding.

CollaborativeEditor

Real-time collaborative AI editor.

CollaborativeSession

Represents a collaborative editing session.

Comparison

Comparison between scenarios

CompilationCache

Compilation cache for statutes.

CompiledRule

Compiled legal rule

ComplianceChecker

Compliance checker

ComplianceMonitoringAgent

Compliance Monitoring Agent.

ComplianceProvider

Provider wrapper that adds compliance and safety features.

ComplianceReport

Compliance report

ComplianceRequirement

Compliance requirement

ComplianceResult

Compliance simulation result

ComplianceSimulator

Regulatory compliance simulator

ComplianceViolation

Compliance violation

Concept

ConceptRelation

ConditionSchema

Schema for a condition in a rule.

ConditionalExecution

Conditional execution rules.

ConfidenceCalibrator

Confidence calibration reporter.

ConfidentialityProtector

Confidentiality protector for filtering sensitive information.

ConflictResolution

Conflict resolution result

ConnectionPool

Connection pool for LLM providers

ConnectionPoolStats

ConsentRecord

ConstitutionalPrompt

Constitutional AI prompting helpers.

Constraint

Constraint between variables

ConstraintProblem

Constraint satisfaction problem

ConstraintSolver

Symbolic constraint solver

ContextAwareSelector

Context-aware prompt selector

ContextCompressor

Context compression for reducing token usage.

ContextImportance

Context importance scorer.

ContextImportanceScorer

Context importance scorer for prioritizing relevant content.

ContextPruner

Automatic context pruner.

ContinualLearner

Continual learning system without catastrophic forgetting.

ContinualLearningStats

ContinualTransferLearner

Continual learner to prevent catastrophic forgetting

ContractAnalysis

Contract analysis result.

ContractClause

Contract clause information.

ContractEvent

Contract event

ContractOptimizationRecommendation

ContractOptimizer

Contract optimization recommender.

ContractReviewAgent

Contract Review Agent.

ContractScenario

Contract scenario

ContractScenarioSimulator

Contract scenario simulator

ContractTerms

Contract terms

Conversation

A conversation with message history and context management.

ConversationConfig

Configuration for conversation management.

ConversationManager

A conversation manager that integrates with an LLM provider.

ConversationOptimizer

Multi-turn conversation optimizer

ConversationTurn

CostAnalytics

Analytics for cost data.

CostAnomaly

Cost anomaly detection result.

CostEstimate

Cost estimate

CostEstimator

Cost estimation for different LLM providers and models.

CostOptimizer

Cost optimizer with recommendations.

CostRecord

Cost record for a single LLM request.

CounterfactualExplanation

CounterfactualGenerator

Counterfactual explanation generator.

CounterfactualResult

Result of counterfactual analysis

CounterfactualScenario

CredentialManager

Secure credential manager

CrossJurisdictionalInfluence

Cross-jurisdictional influence

CrossLingualAnalysis

Cross-lingual legal analysis result

CrossLingualAnalyzer

Cross-lingual legal analyzer

CrossLingualTransfer

Cross-lingual transfer engine

DataSubjectRequest

DatasetBuilder

Dataset preparation utilities.

DatasetStats

Dataset statistics.

DecisionAuditTrail

Decision audit trail.

DecisionRecord

A single decision in an audit trail.

DeepSeekClient

DeepSeek client for advanced reasoning models.

DeploymentBuilder

Builder for LLM deployment specifications.

DeploymentCondition

Condition of a deployment.

DeploymentMetadata

Metadata for LLM deployments.

DeploymentMetrics

Current metrics for a deployment.

DeploymentSpec

Specification for LLM deployment.

DeploymentStatus

Status of an LLM deployment.

DetectedPII

Detected PII in text.

DetectionPattern

Detection pattern for adversarial content

Difference

Document difference

DisclaimerGenerator

Disclaimer generator for legal content

DiskCache

Disk-based cache persistence.

DisputeResolutionAgent

Dispute Resolution Agent.

DistillationConfig

DistillationStats

DistributedInference

Distributed inference coordinator.

DistributedProvider

Distributed provider wrapper that distributes requests across nodes.

DocumentChunk

A chunk of a document with metadata.

DocumentChunker

Document chunker.

DocumentComparator

Document comparator

DocumentComparison

Document comparison result

DocumentQualityScore

Document quality score

DocumentQualityScorer

Document quality scorer

DocumentStructure

Document structure

DocumentStructureAnalyzer

Document structure analyzer

DocumentVersion

DomainAdaptationConfig

Domain adaptation configuration

DomainAdapter

Domain adapter for transferring knowledge between legal specialties

DomainInvariantExtractor

Domain-invariant feature extractor

DomainModel

DomainPromptLibrary

Domain-specific prompt library

DrafterAgent

Drafter agent for creating legal documents.

DynamicPromptTemplate

Dynamic prompt template with variable substitution

EdgeConfig

Edge inference configuration.

EdgeDeployment

Edge deployment manager.

EdgeManifest

Model deployment manifest for edge devices.

EdgeProfile

Edge device profile describing resource constraints.

EdgeProvider

Edge-optimized LLM provider wrapper.

EdgeStats

Edge inference statistics.

EditOperation

An edit operation in a collaborative session.

EffectSchema

Schema for an effect in a rule.

Embedding

A vector embedding representation of text.

EncryptedData

Encrypted data container with metadata

EncryptionService

Simple encryption service (in production, use proper crypto libraries)

Entity

Represents an entity in the legal knowledge graph.

EntityExtractor

Entity extractor for legal documents.

EntityPattern

EthicalBoundaryEnforcer

Ethical boundary enforcer

EthicalCheck

Ethical boundary check result

EthicalRule

Ethical rule

EthicalViolation

Ethical violation

EvaluationBenchmark

Model evaluation benchmark.

EvaluationCase

Single evaluation test case.

EvaluationMetrics

Overall evaluation metrics.

EvaluationResult

Result of evaluating a model on a benchmark.

ExplainableModel

Explainable neuro-symbolic model

Explanation

ExplanationExplorer

Interactive explanation explorer.

ExplanationGenerator

Natural language explanation generator.

ExplanationStyle

ExplanationTemplate

ExplorationStep

ExtractedDate

Extracted date

ExtractedEntities

Extracted entities

FactStore

Memory store for storing and retrieving context information.

FallbackProvider

A provider that tries two providers in sequence, falling back to the second if the first fails.

FeatureAttribution

FeatureAttributor

Feature attribution analyzer.

FederatedConfig

Federated learning configuration

FederatedCoordinator

Federated learning coordinator

FederatedNode

Federated learning participant (e.g., law firm, court, legal department)

FederatedStatistics

Federated learning statistics

FewShotExample

Few-shot learning example

FewShotGenerator

Few-shot learning prompt generator

FewShotLearner

Few-shot learner for new legal domains

FewShotPromptExample

FileStore

File-based conversation store (JSON).

FineTuningDataset

Fine-tuning dataset with train/validation split.

FineTuningExample

Fine-tuning dataset format.

FixtureProvider

Provider that plays back recorded fixtures.

FixtureSet

A collection of response fixtures.

FormField

Form field definition.

Function

A function that can be called by an LLM.

FunctionCall

A function call request from the LLM.

FunctionDefinition

Function definition for API requests.

FunctionOrchestrator

Function orchestrator for advanced execution patterns.

FunctionParameters

Function parameters schema (JSON Schema).

FunctionRegistry

A registry of functions that can be called by LLMs.

FunctionResult

Result of a function execution.

GDPRCompliance

GDPR compliance utilities

GGUFLoader

GGUF model loader for llama.cpp compatible models.

GeminiClient

Google Gemini client.

GeneratedAudio

Represents generated or returned audio in a response.

GeneratedImage

Represents a generated or returned image in a response.

GeneratedPrompt

Generated prompt with metadata

GpuBatchConfig

GPU batching configuration.

GpuDevice

GPU device information.

GpuScheduledProvider

LLM provider with GPU scheduling.

GpuScheduler

GPU scheduler for managing inference requests.

GpuSlot

GPU slot representing an acquired GPU resource.

GrammarGuidedDecoder

Grammar-guided decoder.

GrammarRule

Grammar rule for guided decoding.

GroqClient

Groq client for ultra-fast inference.

GuardedProvider

LLM provider with safety guardrails.

GuardrailsConfig

Guardrails configuration for LLM safety.

Hallucination

Detected hallucination

HallucinationDetection

Hallucination detection result

HallucinationDetector

Hallucination detector

HaystackDocument

Haystack-compatible document format.

Header

Document header

HealthChecker

Health checker for LLM providers.

HealthStats

Health check statistics.

HierarchicalSummarizer

Hierarchical summarization manager.

HistoricalCase

Historical case data

HotSwappableProvider

Hot-swappable provider wrapper.

HuggingFaceClient

HuggingFace Inference API client.

HumanFeedback

HybridInference

Hybrid classical-quantum inference system.

HybridReasoner

Hybrid neuro-symbolic reasoner

HybridSearch

Hybrid search combining semantic and keyword search.

InMemoryDocumentStore

In-memory document store.

InMemoryStore

In-memory conversation store.

IncrementalProcessor

Incremental processor with checkpoint support

InferenceNode

Node information for distributed inference.

InferenceResult

InferenceRule

Inference rule for deriving new relations

InteractiveExplanation

InvalidatableCache

Advanced cache with invalidation strategies.

InvalidatableCacheEntry

Cache entry with invalidation metadata.

JargonNormalizer

Legal jargon normalizer

JsonSchema

A simple JSON schema validator.

JsonSchemaConstraint

JSON schema for constrained generation.

JsonSchemaGenerator

JSON schema-constrained generator.

JudgeBehaviorModel

Judge/jury behavior model

JudicialDecision

Judicial decision

JurisdictionFilter

Jurisdiction-aware safety filter.

JurisdictionInfo

Legal jurisdiction information with language and system metadata

JurisdictionRules

Translation rules specific to a jurisdiction

JurisdictionTransfer

Jurisdiction transfer engine for cross-jurisdictional knowledge transfer

JurisdictionTranslator

Jurisdiction-aware translator

JurisdictionValidationRules

Rules for a specific jurisdiction (validation context)

K8sMetricsCollector

Collects metrics from LLM deployments in Kubernetes.

KnowledgeChange

KnowledgeCompiler

Legal knowledge compiler

KnowledgeDistiller

Legal knowledge distillation system.

KnowledgeExtractor

Automatic knowledge extraction from legal text.

KnowledgeGraphReasoner

Knowledge graph reasoning engine.

KnowledgeGraphStats

KnowledgeSnapshot

LLMConfig

Configuration for LLM requests.

LLMDeployment

Custom Resource Definition for LLM deployments.

LLMOperator

Kubernetes operator for LLM deployments.

LLMSummarizer

LLM-based conversation summarizer.

LangChainMessage

LangChain-compatible message format.

LangChainPromptTemplate

LangChain-compatible prompt template.

LawCompiler

Compiles natural language legal text into structured Statute objects.

LaypersonSummarizer

Layperson-friendly summary generator.

LazyDocumentLoader

Lazy document loader for efficient large document handling

LearningStrategy

LegalAccuracyValidator

Legal accuracy validator

LegalAdversarialProtection

Legal-specific adversarial protection

LegalAnalysisChainBuilder

Builder for creating legal analysis chains.

LegalAnalysisWorkflow

Multi-step legal analysis workflow result.

LegalAnalysisWorkflowPrompter

Multi-step legal analysis workflow prompter (IRAC method).

LegalArgument

Legal argument structure.

LegalCausationAnalyzer

Legal causation analyzer

LegalChange

Detected legal change

LegalChangeDetector

Legal change detection

LegalCitation

Citation extracted from a legal document.

LegalConcept

Legal concept in an ontology

LegalDisclaimer

Legal disclaimer to be added to LLM responses.

LegalDocumentSummarizer

Legal document summarizer.

LegalDocumentSummary

Summary of a legal document.

LegalEntity

Entity extracted from legal text.

LegalEntityExtractor

Legal entity extractor

LegalExpertAgent

Legal expert agent for statute interpretation.

LegalFederatedLearning

Legal domain specific federated learning

LegalFormFiller

Legal form filler using AI.

LegalFormTemplate

Legal form template.

LegalKnowledgeGraph

Legal knowledge graph with entity-relationship mapping.

LegalMetaPrompting

Legal-specific meta-prompting

LegalOntology

Legal ontology

LegalOntologyBuilder

Legal ontology builder for common legal domains

LegalReasoningStep

Chain-of-law reasoning step.

LegalReference

Legal reference

LegalRelation

Relationship between legal concepts

LegalRelationship

Relationship between legal entities or statutes.

LegalResearchAgent

Autonomous Legal Research Agent.

LegalScenario

Legal scenario for what-if analysis

LegalSnapshot

Snapshot of a legal rule at a point in time

LegalTask

Multi-task learning task definition

LibraryStatistics

LitigationRisk

Litigation risk assessment result

LitigationRiskAssessor

Litigation risk assessor

LlamaCppClient

llama.cpp server client for direct local LLM inference.

LlamaIndexDocument

LlamaIndex-compatible document format.

LoRAAdapter

LoRA adapter metadata and weights.

LoRAConfig

LoRA (Low-Rank Adaptation) configuration.

LoadBalancer

Load balancer that distributes requests across multiple providers.

LocalEmbedding

Local embedding provider using sentence transformers via Ollama or similar.

LoggingHook

A simple logging hook for debugging.

LogicGuidedGenerator

Logic-guided neural generator

MemoryAugmentedPrompt

Memory-augmented prompt builder.

MemoryAugmentedProvider

Memory-augmented provider wrapper.

MemoryConfig

Configuration for memory-augmented generation.

MemoryEntry

A memory entry in the memory store.

MemoryItem

MemoryMappedReader

Memory-mapped file reader for large documents

MemoryStore

Memory store for storing and retrieving memories.

Message

A single message in a conversation.

MetaLearner

Meta-learning system for legal domains.

MetaPrompt

Meta-prompt for generating task-specific prompts

MetaPromptingEngine

Meta-prompting engine for automated prompt generation

MetaPromptingStatistics

Statistics about meta-prompting

MetricsCollector

Metrics collector for LLM operations.

MetricsHistory

Training metrics history.

MissingClauseDetector

Missing clause detector

MissingClauseReport

Missing clause report

MistralClient

Mistral AI client.

MockProvider

Mock LLM provider for testing.

ModelLimits

Model token limits for common models.

ModelPricing

Pricing information for a model.

ModelRouter

Model router that selects providers based on routing strategy.

ModelUpdate

Model update from a federated node

ModelVersion

Model version information.

ModerationResult

Moderation result for content.

MonetaryAmount

Monetary amount

MultiModalMessage

A multi-modal message with text and/or images.

MultiModalPromptBuilder

Builder for creating multi-modal prompts easily.

MultiModalResponse

Represents a parsed multi-modal response from an LLM.

MultiTaskLearner

Multi-task learning framework

MultiTenancyManager

Multi-tenancy manager

MultilingualLegalConcept

Legal concept with multilingual support

NegotiationAI

Multi-stakeholder negotiation AI.

NegotiationAssistanceAgent

Negotiation Assistance Agent.

NegotiationParty

Negotiation party

NegotiationRound

Negotiation round

NegotiationSimulation

Negotiation simulation

NegotiationState

NetworkStatistics

Network statistics

NeuralRule

Neural-symbolic rule

NeuroSymbolicExplanation

Explanation for a neuro-symbolic prediction

NormalizationRule

Normalization rule for legal jargon

ObservableProvider

LLM provider with metrics collection.

OllamaClient

Ollama client for local LLM support.

OntologyLearner

Legal concept ontology learner.

OntologyStatistics

Ontology statistics

OpenAIModerationProvider

OpenAI Moderation API integration.

OpenAiClient

OpenAI (or compatible API) client.

OpenAiEmbedding

OpenAI embedding provider.

OptimizationRecommendation

Optimization recommendation.

OptimizationRule

PIIDetector

PII detector using pattern matching.

Paragraph

Document paragraph

ParallelDocumentProcessor

Parallel document processor

ParameterProperty

Property definition for a function parameter.

Participant

Party

Party information

PatternModerator

Basic pattern-based moderator.

Penalty

Penalty

PerformanceCacheWarmer

Performance-oriented cache warmer for preloading common queries

PerformanceMeasuringProvider

Wrapper that measures performance of LLM operations.

PerformanceMetric

PerformanceMetrics

Performance metrics for LLM operations.

PerformanceTimer

Performance timer for tracking operation durations.

PerplexityClient

Perplexity AI client for web-grounded responses.

PrecedentMatch

Legal precedent match result.

PrecedentMatchPrompter

Legal precedent matching prompter.

PrecedentStrength

Precedent strength analysis result

Predicate

Symbolic predicate for logical reasoning

PredictionCandidate

PrivilegeProtection

Privilege protection result

PrivilegeProtector

Attorney-client privilege protector

ProbabilisticLogicProgram

Probabilistic logic program

ProcessingCheckpoint

Checkpoint for incremental processing

ProhibitedActivity

Prohibited activity definition

PrometheusExporter

Prometheus metrics exporter.

PromptAnalysis

Analysis result for a prompt.

PromptAnalytics

PromptChain

A complete prompt chain workflow.

PromptContext

PromptExample

Example for prompt generation

PromptFeedback

PromptMetrics

Metrics for prompt evaluation

PromptOptimizer

Prompt optimization utilities for improving prompt quality.

PromptPerformanceAnalytics

Prompt performance analytics tracker

PromptRefiner

Automatic prompt refiner based on feedback

PromptSelectionStatistics

PromptTemplate

A prompt template with variable substitution.

ProofResult

Result of theorem proving

Proposal

ProviderInfo

Provider capability and cost information.

QualityCriterion

Quality criterion for prompt evaluation

QualityMetrics

Response quality metrics.

QuantizationConfig

Quantization configuration.

QuantizationMetadata

Model quantization metadata.

QuantizationPerformance

Quantization performance metrics.

QuantumAttention

Quantum attention mechanism for legal text analysis.

QuantumBenchmark

Quantum advantage benchmarking system.

QuantumBenchmarkResult

QuantumEmbedding

Quantum embedding generator for legal text.

QuantumGate

QuantumParameters

QuantumSimilaritySearch

Quantum-inspired similarity search for legal documents.

RAGConfig

RAG pipeline configuration.

RAGPipeline

RAG pipeline for retrieval-augmented generation.

RAGStats

Statistics about the RAG system.

RagContextBuilder

RAG-based context builder for legal documents.

RankedClause

RateLimitConfig

Configuration for rate limiting.

RateLimiter

Rate limiter for LLM providers using a token bucket algorithm.

ReActPrompt

ReAct (Reasoning + Acting) pattern builder.

ReRanker

Re-ranker for improving retrieval results.

ReasonerStats

ReasoningChain

Chain-of-thought reasoning trace.

ReasoningFeedback

ReasoningPattern

A legal reasoning pattern that can be learned and improved over time.

ReasoningResult

Result of reasoning

ReasoningTraceVisualizer

Reasoning trace visualizer.

RedlineDocument

Redline document

RedliningEngine

Redlining engine

RefinementRecord

RegressionTestCase

A test case for regression testing.

RegressionTestReport

Report summarizing regression test results.

RegressionTestResult

Result of running a regression test.

RegressionTestSuite

Regression test suite.

Regulation

Regulation

RelationPattern

Relationship

Represents a relationship between two entities.

RelationshipExtractor

Relationship extractor for legal dependencies.

ReplayConfig

RequestBatcher

Request batcher for improved throughput

RequestMetrics

Metrics for a single LLM request.

ResearcherAgent

Researcher agent for finding case law and precedents.

ResourceQuota

ResourceRequirements

Resource requirements for LLM pods.

ResponseCache

A simple LRU cache for LLM responses.

ResponseFixture

A recorded LLM response for fixture-based testing.

RetentionPolicy

RetentionPolicyManager

Data retention policy manager

RetrievedChunk

A retrieved chunk with its relevance score.

RetryConfig

Configuration for retry behavior.

RetryProvider

Wraps an LLM provider with retry logic and exponential backoff.

ReviewerAgent

Reviewer agent for verification and quality control.

RiskFactor

Risk factor

RougeScores

ROUGE (Recall-Oriented Understudy for Gisting Evaluation) metrics.

Rule

Logical rule (Horn clause)

RuleSchema

Schema for a rule extracted from legal text.

SafetyEngine

Safety rules engine.

SafetyRule

Safety rule for custom content validation.

SafetyValidation

Result of safety validation.

ScalingBehavior

Scaling behavior configuration.

ScenarioOutcome

Scenario outcome

ScenarioResult

Scenario simulation result

SchedulerStats

Scheduler statistics.

SearchResult

Section

Document section

SecureCredential

SelfConsistency

Self-consistency decoding implementation.

SelfConsistencyResult

Result from self-consistency decoding.

SelfImprovingReasoner

Self-improving legal reasoning engine.

SemanticCache

Semantic cache for finding similar prompts.

SemanticKernelFunction

Semantic Kernel-compatible function definition.

SemanticKernelParameter

Semantic Kernel parameter definition.

SettlementEstimate

Settlement value estimation result

SettlementEstimator

Settlement value estimator

SimContractClause

Contract clause for simulation

SimplificationMetrics

SimplificationRule

SimplifiedSummary

SlidingWindowContext

Sliding window context manager for long documents.

Stakeholder

Statute

Statute in a specific language

StatuteComparator

Multilingual statute comparator

StatuteComparison

Multilingual statute comparison result

StatuteDifference

Difference between statutes

StatuteInfo

Statute information for validation

StatuteSchema

Schema for a statute structure.

StatuteSchemaGenerator

Statute schema generator.

StatuteSection

Section within a statute.

StatuteSubsection

Subsection within a statute section.

StatutoryInterpretation

Statutory interpretation result.

StatutoryInterpretationPrompter

Statutory interpretation prompter.

StreamChunk

Streaming text chunk from an LLM.

StreamingConfig

StreamingOptimizer

Streaming response optimizer

StudentPattern

SummaryNode

Hierarchical summary node.

SunsetStatute

Statute with sunset clause

SwapStats

Statistics for model swapping.

TableCell

Table cell data.

TableData

Extracted table data.

TabularDataExtractor

Tabular data extractor.

TeacherPattern

TemplateBuilder

Builder for template variables.

TemplateRegistry

A registry for managing multiple prompt templates.

TemporalConflictResolver

Temporal conflict resolver

TemporalKnowledgeTracker

Temporal knowledge evolution tracking.

TemporalLegalRule

Time-aware legal rule

Tenant

Term

TerminologyTranslator

Legal terminology translator

TheoremProver

Automated theorem prover for legal statutes

ThoughtStep

A single step in a chain of thought reasoning process.

TimeoutCancellation

Utility for creating timeout-based cancellation.

TokenEstimator

Simple token estimator.

TokenStats

Statistics for token usage across multiple requests.

TokenTracker

Tracks token usage and provides cost estimation.

TokenUsage

Token usage information for a single request.

TrainingMetrics

Training metrics tracker.

Translation

Translation with metadata

TreeOfThought

Tree-of-thought prompting builder.

UncertainPrediction

UncertaintyEstimate

Uncertainty estimate for a prediction.

UncertaintyQuantifier

Uncertainty quantifier.

UplDetectionResult

UPL detection result.

UplDetector

Unauthorized practice of law (UPL) detector.

UsageStats

ValidatingProvider

Wraps an LLM provider with response validation and retry logic.

ValidationCheck

Individual validation check

ValidationConfig

Configuration for validating provider.

ValidationResult

Validation result with confidence score.

ValidationRule

Validation rule for form fields.

ValidityPeriod

Temporal validity period for a legal rule

ValuationFactor

Valuation factor

VercelAIStreamChunk

Vercel AI SDK-compatible streaming format.

VercelFunctionCall

Vercel AI function call data.

VersionAwareDrafter

Version-aware drafting assistant.

VersionChange

VersionManager

Model version manager for tracking deployed models.

Violation

Violation

VisionConfig

Vision-specific configuration.

WhatIfAnalyzer

What-if analysis engine

WhatIfResult

What-if analysis result

ZeroShotClassifier

Zero-shot legal classifier

ZeroShotLabel

Zero-shot classification label

Enums

AccuracyIssueType

Type of accuracy issue

AgentRole

Agent role in the multi-agent system.

AgentTaskStatus

Agent task status.

AggregationStrategy

Aggregation strategy for federated learning

AnomalySeverity

Severity level for anomalies.

AnonymizationStrategy

ApplicationType

Application type for legal rules

AttackType

Type of adversarial attack

AttributionMethod

AudienceLevel

AudioInput

Represents an audio input in various formats.

AuditResult

BiasCategory

Bias detection categories.

BudgetAlert

Budget alert triggered when thresholds are exceeded.

CaseType

Type of legal case

CausationType

Type of causation in legal analysis

ChangeType

ChunkingStrategy

Strategy for chunking documents.

CircuitState

Circuit breaker states.

ClauseCategory

Clause category

ClauseType

Clause type

CompressionStrategy

Strategy for compressing context.

ConceptType

Type of legal concept

Condition

Types of conditions for conditional execution.

ConditionType

Types of deployment conditions.

ConstraintType

Type of constraint

ContentPart

Represents different types of content in a multi-modal message.

ContextStrategy

Strategy for managing context window when it exceeds limits.

ContinualStrategy

Continual learning strategy

CourtLevel

Court level in judicial hierarchy

DecisionMaker

Decision maker type

DefenseStrategy

Adversarial defense strategy

DifferenceType

Type of statute difference

DisclaimerPosition

Position of the disclaimer in the response.

DisclaimerType

Type of disclaimer

DocChangeType

Document change type

DocDifferenceType

Document difference type

DocSeverity

Document severity level

DocumentType

Document type

EventType

Type of causal event

ExplorationAction

Formula

Logical formula

GateType

GrammarRuleType

Type of grammar rule.

HallucinationType

Type of hallucination

HealthStatus

Health check result.

ImageDetailLevel

Image detail level for vision models.

ImageInput

Represents an image input in various formats.

InfluenceDirection

InvalidationStrategy

Cache invalidation strategy.

JsonType

JSON data types for schema validation.

Jurisdiction

Legal jurisdiction for context-aware analysis.

Language

Language for cross-lingual transfer

LanguageCode

Supported legal language codes (ISO 639-1)

LegalChangeType

Type of legal change

LegalDocumentType

Legal document type for specialized processing.

LegalDomain

Legal domain specialization types

LegalSystemType

Type of legal system

LoadBalancingStrategy

Load balancing strategy for distributed inference.

LogicalOperator

Logical operator for symbolic reasoning

ModerationCategory

Content moderation categories.

NegotiationPhase

NegotiationStrategy

Negotiation strategy

NegotiationStyle

Negotiation style.

OperationType

OptimizationCategory

OptimizationStrategy

Edge optimization strategy.

PIIType

PII (Personally Identifiable Information) type.

ParticipantRole

PartyType

Party type

PenaltyType

Penalty type

Permission

PrivacyMechanism

Privacy mechanism for federated learning

PromptComplexity

Prompt complexity levels.

QuantizationFormat

Quantization format for model weights.

QuantizationPrecision

Quantization precision level.

ReRankingAlgorithm

Re-ranking algorithm for retrieved chunks.

ReadingLevel

ReasoningMethod

Reasoning method used

RecommendationType

Types of optimization recommendations.

RecommendedAction

Recommended action

RedactionStrategy

Strategy for redacting privileged information

ReferenceType

Reference type

RelationType

Type of relationship between legal concepts

RequestStatus

RequestType

ResolutionPreference

Dispute resolution preference.

ResolutionStrategy

Strategy for resolving temporal conflicts

ResultProcessor

Result processing options.

RetrievalStrategy

Strategy for retrieving memories.

RiskLevel

Risk level

Role

Role of a message in a conversation.

RoundOutcome

Round outcome

RoutingStrategy

Routing strategy for selecting providers.

RuleStatus

Status of a legal rule

ScalingMetric

Metrics used for auto-scaling decisions.

SchedulingStrategy

GPU scheduling strategy.

Severity

Severity level

SimplificationCategory

SuggestionStatus

SuggestionType

SwapStatus

Model swap status.

SwapStrategy

Model swap strategy.

TaskComplexity

Task complexity level for routing decisions.

TransferJurisdiction

Jurisdiction for cross-jurisdictional transfer

TreatmentType

How a case treats a cited precedent

UncertaintyType

Types of uncertainty in model predictions.

UplStrictness

Strictness level for UPL detection.

Verbosity

Verdict

Verdict type

VersionChangeType

VisualizationFormat

Visual representation format for reasoning traces.

Vote

Traits

CancellableStreamExt

Extension trait for adding cancellation support to streams.

ConversationStore

Conversation persistence trait.

ConversationSummarizer

Conversation summarizer trait.

DocumentStore

Document store abstraction for RAG.

EmbeddingProvider

Trait for embedding providers.

FunctionHook

Function execution hooks for monitoring and control.

LLMProvider

LLM provider abstraction trait.

ModerationProvider

Content moderation provider.

Functions

bleu_score

Calculates BLEU (Bilingual Evaluation Understudy) score.

calculate_confidence

Calculates a simple confidence score based on response quality indicators.

calculate_quality_metrics

Calculates quality metrics for a text response.

collect_with_cancellation

Helper to collect a stream with cancellation support.

compare_variants

Compares two variants using multiple metrics.

compress_prompt

Prompt compression helper.

merge_tokens

Merges multiple cancellation tokens into a single one. The returned token will be cancelled when any of the input tokens is cancelled.

parallel_map

Parallel map operation over prompts.

rouge_scores

Calculates ROUGE scores.

with_timeout

Wrapper for a stream that supports timeout cancellation.

Type Aliases

BudgetAlertCallback

Callback function for budget alerts.

TextStream

Type alias for stream of text chunks.