HKUDS/RAG-Anything

A graph-powered all-in-one RAG system! RAG-Anything is a graph-driven, all-in-one multimodal document processing RAG system built on LightRAG. It supports all content modalities within a single integrated framework. 100% open-source.

Matched through a repo-specific project phrase query.

A graph-powered all-in-one RAG system! RAG-Anything is a graph-driven, all-in-one multimodal document processing RAG system built on LightRAG. It supports all content modalities within a single integrated framework. 100% open-source.

Matched through a repo-specific project phrase query.

MOST RAG SYSTEMS BREAK THE MOMENT YOU THROW A REAL DOCUMENT AT THEM tables, charts, equations all ignored they only read text... everything else disappears RAG-Anything fixes this multimodal RAG that understands your entire document text, images, tables, formulas and connects them in a knowledge graph. ask a question, get answers from all of it...not just the words github → github.com/HKUDS/RAG-Anythin…

Matched through a repo-specific project phrase query.

This hit 𝟱𝗞+ stars in only 3 months. RAG-Anything is the first RAG that actually handles text, images, tables, and equations in one pipeline. ⚡ Why it matters Academia → Papers need math + diagrams, not just abstracts. Finance → Reports blend commentary, charts, and tables. Tech Docs → Real meaning is in code, diagrams, and data. Enterprise Knowledge → Mixed content demands one pipeline. ⚡ How it works 》 1. Parsing Documents (Ingestion Framework) Every document is broken down with high fidelity→ ✸ MinerU Integration (44K★ GitHub) – Transforms complex PDFs into LLM-ready Markdown/JSON for agent workflows, while preserving document structure. ✸ Adaptive Decomposition – Splits text, visuals, tables, and math while keeping context. ✸ Universal Format Support – PDFs, Office, images, and emerging formats. 》 2. Content Understanding (Processing Layer) Multimodal elements are routed into optimized pipelines→ ✸ Autonomous Categorization – Detects modality type and assigns processing channel. ✸ Concurrent Multi-Pipeline Execution – Runs text + multimodal in parallel. ✸ Hierarchy Extraction – Preserves document order and relationships. 》 3. Multimodal Analysis Engine Dedicated analyzers handle modality-specific logic→ ✸ Visual Content Analyzer – Vision models generate semantic captions + spatial mapping. ✸ Structured Data Interpreter – Finds patterns and semantic links across tables. ✸ Math Parser – Parses LaTeX-level formulas with domain-aware mappings. ✸ Extensible Modality Handler – Plugin-ready for future content types. 》 4. Knowledge Graph Construction Documents are transformed into structured, queryable graphs→ ✸ Entity Extraction – Multimodal elements become graph entities. ✸ Cross-Modal Mapping – Links text with visuals, tables, and math. ✸ Hierarchical Preservation – Maintains original document structure. ✸ Weighted Scoring – Ranks relevance based on context + semantics. 》 5. Intelligent Retrieval (Answer Layer) Hybrid search delivers coherent multimodal answers→ ✸ Vector-Graph Fusion – Semantic embeddings + structural relationships combined. ✸ Modality-Aware Ranking – Adjusts weighting per query type. ✸ Relational Coherence – Keeps retrieved elements contextually aligned. github: github.com/HKUDS/RAG-Anythin… ≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣≣ ⫸ꆛ Want to Build Real-World AI Agents? Join My 𝗛𝗮𝗻𝗱𝘀-𝗼𝗻 𝗔𝗜 𝗔𝗴𝗲𝗻𝘁 𝟱-𝗶𝗻-𝟭 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 — Now with 𝗠𝗖𝗣! ➠ Build Real-World Agents for Healthcare, Finance, Smart Cities ➠ Master 5 Frameworks: MCP · LangGraph · PydanticAI · CrewAI · OpenAI Swarm ➠ Includes 9 Real-World Projects · All Code Included 👉 𝗘𝗻𝗿𝗼𝗹𝗹 𝗡𝗢𝗪 (𝟱𝟬%+ 𝗢𝗙𝗙): maryammiradi.com/ai-agents-m…

Matched through a repo-specific project phrase query.

Try RAG-Anything: github.com/HKUDS/RAG-Anythin…

Matched through a repo-specific project phrase query.

Most RAG systems? They fall apart on real documents. Tables? Ignored. Charts? Skipped. Equations? Gone. They don’t “understand” — they just read plain text. That’s the flaw. RAG-Anything fixes it. Multimodal RAG that actually sees everything: text, images, tables, formulas — all connected. It builds a knowledge graph of your entire document. So when you ask a question — you get answers from everything, not just words. GitHub link in comments 👇

Matched through a repo-specific project phrase query.

👨‍🔧 Github: The all-in-one 100% open-source RAG built on graphs! 3.6K Stars ⭐️ RAG-Anything is a graph-based multimodal RAG for document handling, running on LightRAG. All content modalities are supported in one unified structure. Flexible MinerU-based parsing or direct multimodal content injection workflows Dedicated processors for images, tables, mathematical equations, and heterogeneous content types Advanced search capabilities spanning textual and multimodal content with contextual understanding Automatically identify, categorize, and route different content types through optimized execution channels. --- github .com/HKUDS/RAG-Anything

Contains the exact GitHub repo slug.

MOST RAG SYSTEMS BREAK THE MOMENT YOU THROW A REAL DOCUMENT AT THEM tables, charts, equations all ignored they only read text... everything else disappears RAG-Anything fixes this multimodal RAG that understands your entire document text, images, tables, formulas and connects them in a knowledge graph. ask a question, get answers from all of it...not just the words github → github.com/HKUDS/RAG-Anyt …

Matched through a repo-specific project phrase query.

HKUDS/RAG-Anything: "RAG-Anything: All-in-One RAG System" github.com/HKUDS/RAG-Anythin…

Contains the exact GitHub repo slug.