VectorChat

A Go application that vectorizes text and files into PostgreSQL with pgvector, and uses these vectors to provide context for LLM interactions.

Features

• Vectorize text and files using OpenAI embeddings
• Store vectors in PostgreSQL with pgvector extension
• Retrieve relevant context for LLM queries
• Simple HTTP API for chat interactions

Prerequisites

• Docker and Docker Compose
• OpenAI API key

Setup

1. Clone the repository:

   git clone https://github.com/devs-group/vectorchat.git
   cd vectorchat
   

2. Create a .env file with the required secrets:

   OPENAI_API_KEY=your_openai_api_key
   GITHUB_CLIENT_ID=your_github_oauth_client_id
   GITHUB_CLIENT_SECRET=your_github_oauth_client_secret
   KRATOS_COOKIE_SECRET=at_least_32_random_bytes
   KRATOS_CIPHER_SECRET=at_least_32_random_bytes
   BASE_URL=localhost:4456
   FRONTEND_URL=localhost:3000
   LIGHT_FRONTEND_URL=localhost:3100
   VECTORCHAT_LIGHT_CLIENT_ID=your_vectorchat_light_client_id
   VECTORCHAT_LIGHT_CLIENT_SECRET=your_vectorchat_light_client_secret

   See docs/authentication.md for a full breakdown of the authentication stack and additional environment variables.

3. Start the complete Docker stack:

   docker-compose up --build

4. Access the applications:

   • Main dashboard: http://localhost:3000
   • VectorChat Light: http://localhost:3100
   • API gateway (Oathkeeper): http://localhost:4456
   • Mailhog (dev email): http://localhost:8025

5. To interact with pgAdmin, open http://localhost:5050 and log in with the default email [email protected] and password admin. The database password remains postgres.