Build intelligent
agents in Java
A modular framework with BDI architecture, 55+ semantic annotations, 150+ built-in tools, and multi-agent coordination.
<dependency>
<groupId>io.github.tansuasici</groupId>
<artifactId>tnsai-core</artifactId>
<version>0.2.4</version>
</dependency>Everything you need to build
production-grade AI agents
BDI Architecture
Belief-Desire-Intention models with @RoleSpec, @BeliefSet, and @PlanLibrary annotations for declarative agent design.
Tool Orchestration
SPI-based tool system with ActionExecutor routing across 5 executor types. Register tools via AgentBuilder.tool().
Multi-Agent Coordination
Shared state, message passing, and team patterns with TnsAI.Coordination module.
LLM Streaming
Real-time streaming via streamChatWithHandler with ChatChunk callbacks — TextDelta, ToolCall, Done.
MCP Integration
Full Model Context Protocol support for connecting agents to external tools and data sources.
RAG & Intelligence
Hybrid search, reasoning pipelines, FSM planning, and evaluation benchmarks built in.
Annotation-first design
Define agent behavior declaratively with semantic annotations. No boilerplate, no complex configuration — just clear, readable intent.
Read the docs@AgentSpec(
name = "researcher",
description = "Academic research agent"
)
@RoleSpec(
beliefs = { "research_context" },
desires = { "find_papers" },
capabilities = { "search", "summarize" }
)
public class ResearchAgent extends Agent {
@Tool(description = "Search papers")
public List<Paper> search(String query) {
return rag.hybridSearch(query);
}
}10 modules, one cohesive framework
Ready to build?
Start building intelligent agents with TnsAI today. Open source, Apache 2.0 licensed.