How to Self-Host OpenPencil: Turn Prompts into UI with Docker

OpenPencil turns natural-language prompts into editable vector UI on a live canvas — and because it's fully open source with an official Docker image, you can run your own instance in minutes. This tutorial walks you through a complete self-hosted deployment with the community docker-compose configuration.

Prerequisites

You need a Linux server with Docker and Docker Compose installed. OpenPencil's Rust core is lightweight: 2 vCPU and 4 GB RAM is the comfortable minimum, with 4 vCPU / 8 GB recommended when you run several Agent Teams at once.

Step 1 — Pull the image

The official image is published to GitHub Container Registry:

docker pull ghcr.io/ZSeven-W/openpencil:latest

Step 2 — Run with docker-compose

Create a docker-compose.yml that exposes the web interface on port 8080 and keeps your documents in a persistent volume:

services:
  openpencil:
    image: ghcr.io/ZSeven-W/openpencil:latest
    restart: unless-stopped
    ports:
      - "8080:8080"
    volumes:
      - ./data/openpencil:/data

Then start the stack and verify it is healthy:

docker compose up -d
curl -I http://localhost:8080

Step 3 — Prompt to UI in practice

Open the web client and try the workflow that makes OpenPencil special:

Because every element is real vector geometry, you can tweak any part manually after generation. That hybrid of AI speed and precise human control is exactly what "AI-native" should mean.

Step 4 — Keep it updated

Restart with the latest image regularly to pick up new agents and canvas improvements:

docker compose pull && docker compose up -d

Final thoughts

Self-hosting OpenPencil gives you a private, prompt-driven design studio with no per-seat fees and no cloud dependency. For a full breakdown of requirements, ports, and volume mounts, check the official tool page.