My RAG

My RAG is a retrieval-augmented generation knowledge base system that handles document import, queries, and multi-backend switching.

My RAG is a retrieval-augmented generation knowledge base system featuring one frontend and two equivalent backends built in Python and Java. Both backends share a SQLite database, Milvus vector store, Neo4j knowledge graph, and file storage to let users switch implementations without losing data. It supports document uploads including PDF, Word, Markdown, text, tables, and source code, as well as project directory imports respecting ignore files. The system performs query analysis, multi-path vector retrieval, reranking, and knowledge graph context integration to generate streamed answers with citations and confidence scores. It includes role-based permissions via JWT authentication, session history, feedback collection, and third-party MCP integration.

My RAG at a glance

Best for: Software engineer
Runs on: Web, self-hosted, open source
Pricing: Free
Released: October 2026
Category: AI research tool
Website: github.com/jackwang666666/my-rag

Key features

Knowledge base management with document and directory uploads
Multi-session chat with streaming answers and citations
Knowledge graph extraction and visualization
Role-based permission isolation via JWT
Backend switching between Python and Java implementations
Query analysis and multi-path vector retrieval
Chunk preview and full-text search

Use it for

Answering questions from uploaded knowledge base documents
Comparing Python and Java backend retrieval implementations
Browsing extracted entities and relationships in a knowledge graph
Managing user permissions and document access in a shared system

Alternatives to My RAG

RAG Evaluation Starter — Runs five metrics locally with no SaaS signup or mandatory API key required.
RAG-Anything — It unifies text and multimodal document retrieval within a single framework.
LightRAG — It uses a dual-layer architecture to reduce LLM calls and retrieval latency.
Code Graph RAG — It builds knowledge graphs from codebases using Tree-sitter parsing.
WeKnora — Combines RAG, multi-step agents, and automatic Wiki generation in one platform.
RAGFlow — Provides an integrated platform combining data processing with agent workflows.
LEANN — Efficiently index millions of documents locally with 97% storage savings.
UltraRAG — Build complex RAG workflows using modular MCP components and YAML configuration.
Kotaemon — It offers an open-source, customizable interface for local and API-based document retrieval.
rag_control — Enforces enterprise security policies and tracks RAG execution.
Quivr — It provides a pre-built RAG infrastructure for rapid application integration.
PageIndex — It provides 98.7% accuracy on FinanceBench without vector databases.

All My RAG alternatives