ADVASYS

ADVASYS / ragraph

Public

A local-first, self-organizing AI RAG GRAPH knowledge system that reads, links and reasons over your documents — offline by default, cloud by choice.

12
0
100% credibility
Found Apr 22, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

RAGraph is a desktop app that ingests documents into a hybrid vector-graph knowledge base and powers an autonomous AI agent for chatting with verifiable citations.

How It Works

1
🔍 Discover RAGraph

You hear about a smart app that turns your documents into an organized knowledge web you can chat with.

2
📥 Download and launch

Grab the app and open it on your computer – it runs smoothly like any desktop program.

3
🤖 Connect your AI helper

Pick a smart service like your favorite AI chat to power the thinking and understanding.

4
🌌 Create a knowledge space

Make a new 'universe' for your work notes, research papers, or personal files.

5
📁 Add your folders

Point it to folders with PDFs, docs, or notes – it watches for changes automatically.

6
✨ Watch it organize everything

The app reads, links, and builds a map of ideas, people, and topics from your files.

💬 Chat and explore your knowledge

Ask questions, get precise answers with source links, and browse the living graph of your info.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 12 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is ragraph?

RAGraph is a local-first GitHub Electron app built in TypeScript that ingests your documents—PDFs, DOCX, Markdown, code—from watched folders, chunks them into paragraphs, extracts entities and topics, and builds a self-organizing knowledge graph with typed relations like "works_at" or "part_of". It powers agentic RAG queries where an AI reasons step-by-step over the graph, delivering answers with clickable citations that open exact passages in a source viewer. Offline by default with local embeddings and LLMs like Ollama, but cloud by choice for any OpenAI-compatible provider.

Why is it gaining traction?

Unlike basic vector RAG buckets, RAGraph fuses hybrid search (BM25 + vectors) with graph navigation for structured recall, letting the agent drill from summaries to chunks via tools like entity paths or neighborhood expansion. Users get verifiable, streaming chats with live tool audits, a visual graph browser for exploration, and isolated "universes" for projects—all keeping data in local SQLite/LanceDB files. The self-organizing consolidator links similar docs and merges entities automatically, reducing manual cleanup.

Who should use this?

Knowledge workers indexing research papers or technical docs for Q&A. Developers building personal wikis over codebases or notes. Teams handling sensitive data needing offline RAG without vendor lock-in.

Verdict

Promising local-first RAG graph for private docs, but at 12 stars and 1.0% credibility, it's early—solid docs and tests, yet light on production polish. Try the dev build if you want agentic reasoning over graphs today.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.