# LocalRAG: Yash Ganar

HTML: https://yashganar.com/projects/local-rag
GitHub: https://github.com/yashterdayyyy/LocalRAG
Status: completed · September 2026

## Main aim

Build a RAG system that answers from a dedicated knowledge base. The point is not another chatbot. It is to understand why retrieval still matters when models already offer very large context windows.

A wide window does not replace retrieval. Real systems still hold large, changing, private, and domain-specific knowledge. Pulling only the relevant pieces can improve accuracy, cost, freshness, security, and grounding.

## What shipped

LocalRAG is a full Next.js 16 App Router app: landing page, docs, and a chat UI with sidebar sessions. Markdown under `knowledge/` is ingested into Chroma Cloud collection *secondbrain*. Chat streams grounded answers through Gemini, with hybrid retrieval and MongoDB session history.

Ingest runs via `POST /api/injest`, a re-ingest script, or a GitHub Action that posts changed `knowledge/**` files to the deployed API.

Flow: knowledge/*.md → ingest (API / script / Action) → Chroma Cloud → chat UI → /api/chat → hybrid search + Gemini stream → MongoDB history

## Stack

Next.js 16, React 19, TypeScript, Tailwind, Framer Motion, Chroma Cloud, Gemini embeddings and chat, MongoDB sessions, LangChain text splitters, Vercel AI SDK, and GitHub Actions for knowledge ingest.

Highlight icons on the site: MongoDB, Chroma, Gemini, React.

## Repository

Source: https://github.com/yashterdayyyy/LocalRAG