Deployment | VexAI Docs

Docker with SQLite (Recommended)

The fastest way to get VexAI running in production. SQLite requires zero database configuration.

git clone https://github.com/mrelmida/vexai.git
cd vexai
cp .env.example .env
# Edit .env with your Discord token + LLM API key

docker-compose up -d --build
docker-compose logs -f

docker-compose.yml

services:
  vexai:
    build: .
    restart: unless-stopped
    env_file: .env
    volumes:
      - ./data:/app/data
      - ./skills:/app/skills
    ports:
      - "3847:3847"  # webhook server (if enabled)

💡 Tip

The data/ volume persists your database, config, and identity files across container restarts and upgrades.

Docker with PostgreSQL + pgvector

For production deployments that need advanced full-text search and native vector embeddings.

docker-compose -f docker-compose.postgres.yml up -d --build
docker-compose -f docker-compose.postgres.yml logs -f

This compose file includes a PostgreSQL + pgvector container alongside the bot, pre-configured with connection pooling and automatic schema migration.

ℹ️ Info

The PostgreSQL compose file automatically sets DATABASE_TYPE=postgres and configures the connection URL. No manual database setup required.

Native Node.js

Run VexAI directly on your machine without containerization.

Prerequisites

Node.js v22+: required for ESM and modern API support
npm: ships with Node.js

npm install
npm run build
npm start

# Development with hot-reload:
npm run dev

Build Commands

Command	Description
`npm run build`	Compile TypeScript → `dist/`
`npm run dev`	Run with hot-reload (tsx watch)
`npm start`	Run compiled `dist/index.js`
`npm run typecheck`	Type check without emitting (`npx tsc --noEmit`)

Project Structure

src/
  bot/            # Core Discord client, message handling
  agents/         # Specialized agent system
  sub-agents/     # Background agent execution
  workflows/      # Workflow automation engine
  events/         # Event-driven reaction system
  llm/            # Multi-provider LLM abstraction
  skills/         # Modular skill framework (22 built-in)
  integrations/   # External service connectors
  voice/          # Voice and audio processing
  security/       # Security observer and access control
  cache/          # Performance caching layer
  database/       # Database abstraction (SQLite + PostgreSQL)
  identity/       # Bot personality and memory files
  config/         # Zod-validated configuration
  logging/        # Winston structured logging
  utils/          # Shared utilities
  types/          # TypeScript declarations
data/             # Runtime data (database, identity, user data)
skills/           # User-created hot-loadable skills

Production Tips

Use Docker: consistent environment, automatic restarts, and simple upgrades
Enable PostgreSQL: recommended for high-traffic servers with concurrent users
Cheaper observer model: configure a separate, less expensive model for the security observer to reduce costs
Reverse proxy: place the webhook server behind nginx or Caddy with HTTPS termination
Monitor logs: VexAI uses Winston structured logging; pipe to your preferred log aggregator

⚠️ Warning

Always back up your data/ directory before updates. It contains your database, configuration, identity files, and user memory.

🐳 Deployment