Rebuff.AI

Tooling

Frontend:
- Next.js - a React framework for production.
- Tailwind for styling and layout.
- Supabase.js for user management and realtime data syncing.
Backend:
- app.supabase.com: hosted Postgres database with restful API for usage with Supabase.js.

Database Setup

In order to use the Rebuff server, you'll first need to:

Create relational database tables and functions to record user's credits and previous attempts
Create a vectorDB to record previous prompt embeddings for similarity searching

Supabase (Relational DB)

The needed relational database tables can be found in sql_setup/tables
The needed relational database functions can be found in sql_setup/functions

Vector DB

The vector DB can be either Pinecone or Chroma.

Pinecone

The pinecone index that you create must be of dimension 1536.

Chroma

To set up Chroma, follow the deployment instructions.
When you start the Rebuff server, you will need to set the following environment variables:
- CHROMA_URL: The HTTP URL of the Chroma server (e.g. http://localhost:8000)
- CHROMA_COLLECTION_NAME: The collection to use. This will be created if it does not exist.
- VECTOR_DB: Set this to chroma.
At the current time, authentication to the Chroma server is not supported, so make sure the server is adequately secured.