TnsAI Core 0.18.3 API
Packages
Package
Description
TnsAI Java SDK - A multi-agent framework with BDI architecture and Gaia methodology.
Action execution framework for agent tool/function calling.
Design-by-contract support for action execution with pre/post condition validation.
Specialized action executors for different execution contexts.
Typed models for action execution.
Agent classes and lifecycle management for the TnsAI framework.
Agent annotations for declarative agent definition.
Asynchronous agent execution with CompletableFuture, reactive streaming, and session management.
Agent capability management -- planning, reasoning, evaluation, feedback,
environment perception, variant selection, context graphs, and resilience.
Agent chat orchestration for LLM invocation, history management, and tool call handling.
BDI Cognitive Engine components for agent reasoning.
Multi-LLM ensemble execution with parallel fan-out, race patterns, and consensus voting.
Agent execution components including LLM orchestration, tool processing, and memory management.
Agent group membership management.
Hierarchical agent relationships with parent-child management and task delegation.
Agent initialization and specification extraction.
Agent interruption and checkpoint system.
Agent lifecycle management with annotation-based hooks, timeout enforcement, and idle monitoring.
Inter-agent messaging for task delegation, query handling, and protocol communication.
Agent orchestration — chat, streaming, tool execution, and knowledge base management.
Agent output formatting and serialization to various formats (JSON, YAML, XML).
Agent resilience management components.
Streaming and event-based chat operations with token-level granularity and event publishing.
Agent variants with runtime switching and automatic selection based on task complexity.
Custom annotations for the TnsAI framework.
Reusable action-dispatch contracts for
Roles.Agent messaging infrastructure for inter-agent communication.
Message transport layer for agent communication.
Configuration management and environment variable loading.
Context graph infrastructure for decision tracking and auditing.
Context assembly and token estimation for LLM interactions.
Decision tracking and auditing for agent context graphs.
Entity resolution for multi-source context graphs.
Context snapshots capturing agent state at specific moments in time.
Shared enumerations used across the TnsAI framework.
Agent-related enumerations for personality and behavior configuration.
Role-related enumerations for parameter and safety classification.
Environment and world model abstractions for BDI agents.
Evaluation infrastructure for agent output quality assessment.
Evaluation hooks for agent lifecycle integration.
Event-driven communication system for agent lifecycle and execution notifications.
Runtime processing infrastructure for
EventHandler annotations.Custom exceptions for the TnsAI framework.
Runtime enforcement of
InputGuardrail and
OutputGuardrail annotations declared on
@ActionSpec methods or their declaring Role classes.Decentralized identity management for agents using W3C DID standards.
Invariant checking framework for runtime state verification.
Knowledge base abstraction for Retrieval-Augmented Generation (RAG).
LLM (Large Language Model) client interfaces and abstractions.
Memory management and persistence for agent conversation history.
Advanced memory module: vector search, BM25 text search, Reciprocal Rank
Fusion hybrid retrieval, skill memory, importance scoring, staleness
detection, and session persistence.
Memory management strategies for conversation history within token budgets.
Metadata extraction and representation for actions, tools, and roles.
Data models for the TnsAI framework.
BDI (Belief-Desire-Intention) model classes for agent reasoning.
Multimodal content parts for LLM communication.
Gaia methodology model classes for role-based agent design.
Structured error reporting infrastructure.
Observability event schema + thread-local correlation context.
Output serialization framework for converting action results to structured formats.
Annotations for declarative output format configuration.
Concrete output serializer implementations for standard formats.
LLM output parsing with retry and validation support.
Prompt building, templating, and management framework for LLM interactions.
Predefined prompting strategies for enhanced LLM response quality.
SPI handles for agent protocol management.
Declarative annotations for configuring agent communication protocols.
Protocol callback handlers for annotation-driven message dispatch.
Protocol data models for inter-agent message exchange.
Protocol transport layer for message delivery between agents.
Canonical RAG configuration and retrieval contracts.
Resilience patterns for fault-tolerant agent and action execution.
Dead letter queue for capturing and replaying failed agent messages.
Agent failure detection, health tracking, and automated recovery.
Role definitions and capabilities for agents in the TnsAI framework.
Role annotations for declarative role definition.
Role export system for converting TnsAI roles to various formats.
Framework-specific role exporters.
Role metadata extraction from annotations and class definitions.
Centralized role registration and lookup service.
JSON Schema generation for LLM function calling tool definitions.
Security SPI contracts and human-in-the-loop approval for TnsAI agents.
Human-in-the-loop (HITL) approval mechanism for sensitive agent actions.
Persistent audit trail for security-relevant decisions.
Service Provider Interface (SPI) package for TnsAI.
State persistence and checkpointing for agent recovery.
Tool framework for defining, registering, and discovering agent capabilities.
Centralized metrics collection and reporting for tool executions.
Service Provider Interface for pluggable tool implementations.
General-purpose utility classes shared across the TnsAI framework.