RAG Platform

A minimal RAG (Retrieval-Augmented Generation) app. Upload documents into named pipelines and chat with an AI agent grounded on your content.

Features

• Create multiple document pipelines (knowledge bases)
• Upload PDF, TXT, and Markdown files
• Chat with retrieved context via Ollama (default, local & free)
• Optional bring-your-own-key for OpenAI or Anthropic
• Protected by a shared app API key (no user accounts)

Prerequisites

• Python 3.11+
• Node.js 18+
• Ollama running locally (or via Docker)

Pull required Ollama models:

ollama pull nomic-embed-text
ollama pull llama3.2

Quick Start (local)

One command (starts Ollama if installed, backend, and frontend):

./start.sh

To stop backend and frontend:

./start.sh stop

Or start manually:

1. Copy environment config:

cp .env.example .env

2. Start the backend:

cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

3. Start the frontend (new terminal):

cd frontend
npm install
npm run dev

4. Open http://localhost:5173 and enter the API key from .env (change-me by default).

Docker

cp .env.example .env
docker compose up --build

After Ollama starts, pull models inside the container:

docker compose exec ollama ollama pull nomic-embed-text
docker compose exec ollama ollama pull llama3.2

• Frontend: http://localhost:5173
• Backend API: http://localhost:8000
• API docs: http://localhost:8000/docs

Usage

1. Enter the app API key on first visit.
2. Create a pipeline (e.g. "Q3 Reports").
3. Upload documents — they are parsed, chunked, embedded, and stored in Chroma.
4. Ask questions in the chat panel. Answers cite source filenames.
5. Switch provider to OpenAI or Anthropic and enter your API key for cloud models.

API

Method	Endpoint	Description
GET	/health	Health check + Ollama status
POST	/pipelines	Create pipeline
GET	/pipelines	List pipelines
DELETE	/pipelines/{id}	Delete pipeline
POST	/pipelines/{id}/documents	Upload document
GET	/pipelines/{id}/documents	List documents
POST	/pipelines/{id}/chat	RAG chat (SSE stream)

All endpoints except /health require the X-API-Key header.

Project Structure

backend/     FastAPI API, ingest, retrieval, LLM
frontend/    Vite + React minimal UI
db/          SQLite metadata, Chroma vectors, uploads (gitignored)