Unbody Backend Starter

A TypeScript-based backend template for Unbody integration, providing AI-powered content management capabilities.

Project Structure

.
├── backend/                 # Backend services and configurations
│   ├── admin-client.ts     # Unbody admin client configuration
│   ├── configs.ts          # Project configurations and constants
│   ├── run.ts              # Script runner for backend operations
│   ├── setup-project.ts    # Project initialization and setup
│   ├── enhancers/          # Custom content enhancers
│   └── data-sources/       # Data source configurations
└── package.json           # Project dependencies and scripts

Prerequisites

• Node.js (v14 or higher)
• Yarn package manager
• Unbody account and credentials

Environment Variables

Create a .env file in the root directory with the following variables:

UNBODY_ADMIN_ID=your_admin_id
UNBODY_ADMIN_SECRET=your_admin_secret

Installation

1. Clone the repository
2. Install dependencies:

   yarn install

Unbody Integration

One-Time Setup Routines

These commands need to be run only once when setting up the project:

1. Initialize the Unbody project:

   yarn unbody-backend:setup-project

   This creates a new project in Unbody with configured AI services and settings.

2. Create data sources:

   yarn unbody-backend:data-sources:push-datasource-create

   This sets up the data sources that will be used to push content to Unbody.

Recurring Data Update Routines

These commands should be run whenever your local data changes:

1. Update data sources with new content:

   yarn unbody-backend:data-sources:push-datasource-update

   This pushes any new or updated content to Unbody.

Backend Configuration

The backend is configured with the following AI services:

• Text Vectorizer: OpenAI Text Embedding 3 Large
• Generative Model: OpenAI GPT-4
• Reranker: Cohere English V3
• Auto Summary: OpenAI GPT-4 Mini
• Auto Vision: OpenAI GPT-4

Project Settings

The project is configured with:

• Project Name: the name of Unbody project
• Storage Path: where the files are stored

Supported file extensions:

• PDF, DOCX, DOC, TXT, MD, CSV, XLS, XLSX

Development Workflow

1. For new projects, run the one-time setup routines in order:

   yarn unbody-backend:setup-project
   yarn unbody-backend:data-sources:push-datasource-create

2. When your local data changes, run the update routine:

   yarn unbody-backend:data-sources:push-datasource-update

License

This project is licensed under the MIT License.