Ragent AI is a Java-based enterprise Agentic RAG platform covering documentation ingestion, multi-channel retrieval, intent recognition, MCP tool calls, streaming responses, and end-to-end tracing. It emphasizes production-grade architecture with modular Maven modules and multi-channel retrieval.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Ragent AI is a production-grade Java AI application platform focused on Agentic RAG. It covers from document ingestion to streaming answers, multi-channel retrieval (vector, keyword, knowledge graph, and internet search), intent recognition, model routing and MCP tool calls, streaming responses, and end-to-end tracing. The project is organized into four Maven modules: framework, infra-ai, bootstrap, and mcp-server.
How it works
Key design elements include: modular Maven-based backend with clear boundaries between infrastructure, AI capabilities, RAG business logic, and MCP services; multi-channel retrieval with parallel execution and post-processing (de-dup, RRF fusion, rerank, metadata enrichment); model routing with fault tolerance via a three-state circuit breaker; a node-based ingestion pipeline with configurable tasks and per-node logs; and extensible interfaces for models, storage, retrieval, ingestion, and MCP tools. The system emphasizes separation of concerns so switching models, vector stores, or storage implementations does not require rewriting the main QA flow.
Getting started
The repository provides a production-ready Java AI application. It includes documentation and a React-based console for user and admin access. The latest release is 1.0.0 (June 16, 2026) marking production-ready stability with end-to-end validation from document ingestion to streaming QA.
Recent releases
Latest: 1.0.0 (June 16, 2026) — Stable first production release. This release marks the transition from early preview to production-ready, with end-to-end verification from document ingestion to streaming question answering.
Traction
Stars: 3432; Forks: 695
Behind the repo
Not provided in the README excerpt.
Caveats
License: Apache-2.0; Language: Java; Created: 2025-11-20; Last push: 2026-08-04. Observed production-oriented features include concurrency controls, circuit breakers, distributed tracing, and data consistency guarantees.