rag-local

Your private memory for AI assistants

Your AI assistant AI rag-local local Docs
Want it as a serverless API? Deploy it to Cloudflare Pages. See the Cloudflare Pages mode →

What is it?

A simple way to give your AI assistant access to your own notes and documents, without anything leaving your computer.

Private

Your documents never leave your machine. No external servers, no cloud.

Yours forever

Everything is stored in files you back up and control. Your knowledge, your rules.

Works with any assistant

Connects with Claude and other compatible assistants, so they answer using your data.

How it works

1

You save your notes

You create collections and add your documents using a simple format.

2

The system understands them

rag-local reads and organizes the content so it can be searched meaningfully.

3

Your assistant queries them

When you ask something, your assistant searches your collections and answers with your data.

It's like having a personal library with its own librarian.

Installation

1Install Node.js

Download and install Node.js from nodejs.org.

2Download the project

Clone the repository or use Download ZIP on GitHub.

3Open the terminal and paste two commands

In the project folder, open a terminal and paste these two commands, one at a time:

cd rag-node && npm install
node rag-server.mjs

4Open the local web page

In your browser, go to http://localhost:8937.

The first time, rag-local downloads about 316 MB.

Everyday use

Create a collection

You build a collection with a name and add documents using a simple format (OKF): title, description, and tags.

Title: Bread recipes
Description: How to make homemade bread step by step.
Tags: cooking, bread, easy

Search

You type a question in plain language and the system shows you the most relevant snippets, with a score showing how useful each one is.

Back up

You export a collection as a single portable file (.jvsb) and save it wherever you want.

Connect it to your assistant

So Claude can use your collections, run this single command in the terminal:

claude mcp add rag-local -- npx -y @rckflr/mcpwasm http://localhost:8937

That lets Claude read your collections to answer, and also create new ones from your conversations.

Frequently asked questions

Does my data leave my computer?

No, never. All processing happens on your machine.

Do I need to know how to code?

Not to use it. Just copy three commands.

How much disk space does it use?

About 350 MB including the AI model.

Does it work without internet?

Yes. After the first model download, rag-local runs 100% offline.

Can I delete everything?

Yes. Delete the project folder and nothing is left behind.