Skip to content

Quickstart

Get MIFY running locally and execute your first workflow.

SoftwareVersionPurpose
Node.js22+JavaScript runtime
pnpm9+Package manager
PostgreSQL15+Database (with pgvector extension)
GitAnyVersion control

Optional:

  • Ollama — for local AI models (no API key needed)
  • Redis — for caching and rate limiting
Terminal window
git clone https://github.com/HYGO-INC/mify.git
cd mify
pnpm install
Terminal window
# Create a PostgreSQL database
createdb mify
# Enable pgvector extension
psql mify -c "CREATE EXTENSION IF NOT EXISTS vector;"
# Run migrations
pnpm --filter @mify/api prisma:migrate dev

Copy the example environment file and set your database URL:

Terminal window
cp .env.example .env

Edit .env:

Terminal window
DATABASE_URL=postgresql://user:password@localhost:5432/mify
Terminal window
# Start the API server (port 3010)
pnpm --filter @mify/api dev
# In a separate terminal, start the web UI (port 3006)
pnpm --filter web dev

Navigate to http://localhost:3006 in your browser.

Click Canvas in the navigation to open the visual workflow editor.

Every workflow starts with a trigger. Drag a Manual Trigger node onto the canvas — this lets you run the workflow by clicking a button.

Drag a Chat node onto the canvas. This sends a prompt to an AI model and returns the response.

Configure it:

  • Provider: Select your AI provider (Ollama for local, or OpenAI/Anthropic with an API key)
  • Model: Choose a model (e.g., phi3:mini for Ollama, gpt-4o for OpenAI)
  • System Prompt: “You are a helpful assistant.”

Click the output handle of the Manual Trigger and drag to the input handle of the Chat node.

Click Run in the toolbar. Enter a message like “What is MIFY?” and see the AI respond.

ServiceURLExpected
API Serverhttp://localhost:3010/health{ "status": "ok" }
Web UIhttp://localhost:3006MIFY interface loads