All Classes and Interfaces
Class
Description
Handles action execution in a ReAct loop.
Enriches system prompts with learned rules from
FeedbackLearner.Executes LLM-generated plans with adaptive replanning on failure.
Builder for
AdaptiveReplanEngine.Generates detailed agent configurations (system prompts and tool assignments)
from a
TeamPlan.Manages context graphs for an agent.
Represents a complete agent session with all conversation turns.
Aggregation strategies for combining multiple reasoning paths.
Main processor for Atom of Thought (AoT) reasoning.
Result of Atom of Thought (AoT) processing.
Represents an atomic reasoning unit in Atom of Thought (AoT) processing.
Atom processing status.
Atom types for categorization.
Generates atomic reasoning units from complex problems.
Solves atoms in parallel with dependency resolution.
Synthesizes atom solutions into a final coherent answer.
Synthesis strategies.
Monitors conversation state and triggers consolidation automatically.
Generates a fully configured multi-agent team from a natural language task description.
Result of the auto-team building process.
Default planner implementation using backward chaining algorithm.
Maps BDI (Belief-Desire-Intention) model to Context Graph components.
Evaluates branches in Tree of Thoughts exploration.
Built-in criteria checks for the refinement loop.
Describes an available capability (tool or action) for the LLM planner.
Types of causal queries.
Causal reasoning engine for why, what-if, and intervention queries.
Result of a causal reasoning query.
Checkpointer-backed implementation of DecisionTraceStore.
Checkpointer-backed implementation of SnapshotStore.
Configuration for context compaction.
Builder for CompactionConfig.
Result of a context compaction operation.
Defines completion criteria for the refinement loop.
Builder for CompletionCriteria.
Evaluates condition expressions against a state (belief) map.
A consideration that contributes to an action's utility score.
Result of a self-consistency reasoning execution.
Builder for ConsistencyResult.
Result of a memory consolidation operation.
When to trigger memory consolidation.
Interface for context compaction strategies.
Detects contradictions between incoming and existing temporal knowledge.
Strategy for checking whether two facts conflict with each other.
A detected contradiction between two temporal knowledge items.
Represents a single turn in an agent conversation.
Interface for criteria checks in the refinement loop.
Default factory for creating
AgentContextManager instances.Default
FeedbackCollector.Factory implementation.Default factory for creating
PlannerHandle instances.Default evaluator using keyword matching and similarity metrics.
Default factory for creating
ReasoningStrategyHandle instances by name.Manages ephemeral tool outputs to prevent context explosion.
Configuration for an ephemeral action.
Tracks a single tool result.
Contract for thought nodes that can be evaluated by a
BranchEvaluator.Represents an event that triggers state transitions.
Extended thinking mode client for deep reasoning tasks.
Builder for ExtendedThinkingClient.
Represents knowledge extracted from a session.
Builder for ExtractedKnowledge.
Types of knowledge that can be extracted from sessions.
Represents user or system feedback on agent output.
Learns from user feedback to improve agent behavior over time.
Result of a learning cycle.
Storage for agent feedback records.
Types of feedback that can be provided on agent output.
File-based implementation of KnowledgeStore.
File-based implementation of SessionStore.
Memory decay strategies based on forgetting curves.
Configuration for a single agent within a generated team.
A node in a Graph of Thoughts.
Operations available in Graph of Thoughts.
Result of a Graph of Thoughts exploration.
Graph of Thoughts (GoT) executor for complex problem solving.
Functional interface for transition guard conditions.
In-memory implementation of DecisionTraceStore for development and testing.
In-memory implementation of EntityResolver for development and testing.
In-memory implementation of KnowledgeStore.
In-memory implementation of SnapshotStore for development and testing.
Extracts knowledge from agent sessions.
Filter for searching extracted knowledge.
Builder for KnowledgeFilter.
Store for extracted knowledge.
Statistics about stored knowledge.
Strategies for learning from feedback.
LLM-based context compactor that summarizes old messages.
LLM-powered dynamic planner that uses a language model to decompose
natural-language goals into executable step sequences.
Builder for
LLMDynamicPlanner.LLM-based knowledge extractor.
Utility for parsing structured content from LLM responses.
An LLM-generated execution plan.
A single step in an LLM-generated plan.
Automated memory consolidation pipeline.
Runtime norm evaluation and enforcement engine.
Result of checking an action against norms.
Runtime representation of a norm extracted from annotations.
Represents a violation of a norm.
Result of prompt optimization.
Tracks progress of a prompt optimization session.
Record of a single optimization iteration.
Record of a strategy application.
Plan-level human-in-the-loop approval gate.
Represents the user's decision after reviewing a plan.
Type of plan approval decision.
Builder for
PlanApprovalGate.Callback for presenting a plan to the user for review.
Result of executing an entire LLM-generated plan.
Executes plans generated by a
Planner on a role instance.Service Provider Interface for planning algorithms.
Runtime representation of a planning action.
Builder for PlanningAction.
Typed failure raised when planning cannot produce a plan for an unsatisfied
goal, or when a produced plan stops holding while it executes.
Why planning ended without a plan.
Runtime representation of a planning goal.
Parser for PRD (Product Requirements Document) JSON files.
Persistent learning store for cross-task knowledge transfer.
A recorded learning from task execution.
Evaluates prompt responses against expected outputs.
Loads prompt templates from files and resources.
Automated prompt optimization framework.
Central registry for prompt templates with hot-reload support.
Pruning strategies for Tree of Thoughts exploration.
ReAct (Reason + Act + Observe) executor for agentic reasoning.
Builder for ReActExecutor.
Result of a ReAct (Reason + Act + Observe) execution.
Status of a ReAct execution.
Represents a single step in a ReAct (Reason + Act + Observe) loop.
Represents a single reasoning path in self-consistency or tree-of-thoughts execution.
Builder for ReasoningPath.
Routes queries to reasoning-specialized models based on task complexity.
Builder for ReasoningRouter.
Verifies reasoning chains for logical consistency and factual accuracy.
Iterative refinement loop for persistent output improvement.
Builder for RefinementLoop.
Result of a refinement loop execution.
Status of refinement completion.
Represents a single step in a session replay.
File-based session store with search capabilities.
Extended session store with search capabilities.
Session statistics.
Self-Consistency reasoning executor.
Builder for SelfConsistencyExecutor.
Outcome of an agent session.
Query builder for searching sessions.
Builder for SessionQuery.
Search mode for text queries.
Provides replay functionality for agent sessions.
Statistics about a session replay.
Result of a session search operation.
A matched turn within a session.
Position of a match within content.
Interface for storing and retrieving agent sessions.
Simple prompt template implementation using the SimpleTemplateEngine.
Builder for SimplePromptTemplate.
Simple template engine with {{variable}} syntax.
Represents a state in a Finite State Machine.
Builder for State.
Functional interface for state entry/exit actions.
A Finite State Machine implementation for bounded agent autonomy.
Builder for StateMachine.
State machine status.
Context object passed to state actions and guards.
Record of a state transition.
Exception thrown when state machine operations fail.
Listener interface for state machine events.
Generates visual representations of state machines.
Options for Graphviz DOT generation.
Options for Mermaid diagram generation.
Result of executing a single plan step.
Functional interface for executing a single plan step.
Determines when a ReAct loop should stop.
Mutable context for stop conditions that need to track state across steps.
Stop hooks for controlling when the refinement loop should terminate.
Ensures LLM outputs conform to expected schema with automatic retry.
Result of a structured output generation attempt.
Generates synthetic test data for prompt optimization.
Configuration for test case generation.
Uses LLM to decompose a natural-language task into agent roles and capabilities.
A single task in a task list.
Builder for TaskItem.
An ordered list of tasks to execute.
Executes a task list using iterative refinement for each task.
Builder for TaskListExecutor.
Result of executing a complete task list.
Execution status of a task item.
Intermediate representation of a generated team plan.
A fact with temporal validity boundaries.
Test case for prompt optimization.
Result from extended thinking mode execution.
Builder for ThinkingResult.
Format for the thought phase of ReAct.
Represents a node in a Tree of Thoughts exploration.
Builder for ThoughtNode.
Rough token count estimator for text content.
Analyzes inter-agent dependencies and selects the optimal coordination formation.
Supported formation types.
Result of a Tree of Thoughts exploration.
Builder for ToTResult.
Represents a transition between states in a Finite State Machine.
Full builder for Transition.
Builder step for specifying event.
Builder step for specifying target state.
Functional interface for transition actions.
Tree of Thoughts (ToT) executor for complex problem solving.
Builder for TreeOfThoughtsExecutor.
Phase-1 compactor that truncates large tool-call arguments in older messages.
Types of conversation turns.
Two-phase context compactor combining truncation and LLM summarization.
Utility AI planner that selects actions based on cost/value trade-offs.
Builder for UtilityAIPlanner.
Result of a validation check.
A validation rule for structured output fields.
Result of reasoning verification.
Builder for VerificationResult.