Yuliu11

Yuliu11 / DualStack-Agent

Public

Agentic-RAG Framework: A high-performance knowledge retrieval system featuring Hybrid Search (FAISS + BM25), RRF Re-ranking, and MySQL/Redis persistent storage.

64
0
100% credibility
Found Feb 26, 2026 at 33 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

An open-source tool for processing PDF documents into a smart question-answering system that provides answers with source citations using hybrid search techniques.

How It Works

1
📖 Discover the document chat tool

You find a helpful app that turns your PDF reports into a smart conversation partner.

2
🗂️ Gather your files

Collect your PDF documents, like company reports, and place them in a folder.

3
🚀 Wake up the assistant

Hit start, and your personal document expert springs to life on your screen.

4
📥 Share your documents

The assistant eagerly reads and learns from every page in your files.

5
💬 Ask away

Chat naturally, like 'How did sales change last year?', and watch magic happen.

✅ Unlock clear answers

Receive spot-on responses with direct quotes from your docs, building total trust.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 33 to 64 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 DualStack-Agent?

DualStack-Agent is an agentic RAG framework in Python that ingests PDFs like annual reports, parses text and tables, and builds a searchable knowledge base with hybrid FAISS vector + BM25 keyword search, fused via RRF re-ranking. It stores everything persistently in MySQL with Redis caching, exposes a FastAPI chat endpoint for queries, and includes streaming responses plus a simple Vue chat UI. Developers get agentic smarts like intent routing (fact, calc, summary) and auto-calculations for metrics like growth rates, solving unreliable RAG on structured docs.

Why is it gaining traction?

It stands out as an agentic rag framework github project with production hooks: idempotent API calls prevent duplicate LLM costs under concurrency, dual-LLM fallback (DeepSeek + DashScope) with circuit breakers, and 30% faster multi-route retrieval via async. Users notice accurate financial comparisons across docs without hallucinations, thanks to calc tools and citations, plus easy Docker-local setup for MySQL/Redis.

Who should use this?

Backend devs prototyping agentic AI RAG for financial QA, like comparing revenues in annual reports or building internal knowledge agents. Teams evaluating open source agentic rag frameworks for PDF-heavy workflows, especially those needing persistent storage and hybrid search over pure vector setups.

Verdict

Low 1.0% credibility score and 16 stars signal early-stage maturity—docs are solid with API examples and Docker compose, but expect tweaks for scale. Worth forking for agentic rag github experiments if you need calc-aware RAG now.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.