Quickstart (Self-Host)

Get Deckyard running on your machine. The installer takes one line; the manual path takes five.

What This Is

Deckyard is a self-hosted web app for creating and presenting slide decks. It runs as:

A plain Node.js server
A vanilla ESM client (no bundler)
File-based JSON persistence on disk, with Postgres as an optional mode

Requirements

Either of these is enough; the installer uses whichever it finds.

Docker, or
Node.js 22 or newer with npm

The installer does not install Docker or Node for you. If neither is present it says where to get them and stops.

One-Line Install

curl -fsSL https://deckyard.eu/install.sh | bash

This clones the repository, auto-detects Docker or Node, writes a local .env, installs dependencies, starts the app and opens your browser at http://localhost:4177. Nothing leaves your machine.

It is safe to re-run: an existing install is updated in place, and an existing .env is left alone.

Optional knobs, all set as environment variables:

Variable	Effect
`DECKYARD_DIR`	Target directory (default: `./deckyard`)
`DECKYARD_BRANCH`	Branch to check out (default: `main`)
`DECKYARD_MODE`	Force `docker` or `node` instead of auto-detecting
`PORT`	App port (default: `4177`)

Piping a script into bash deserves a look first. It is served at deckyard.eu/install.sh, and the same file is scripts/install.sh in the repository.

Manual Install

git clone https://github.com/jaapstronks/deckyard.git
cd deckyard
npm install
npm run setup   # optional: AI key, auth, port (Enter accepts defaults)
npm run start

Access

Open your browser to: http://localhost:4177

Configuration

npm run setup writes a .env for you. To edit it by hand, copy .env.example to .env and change what you need.

See Environment Variables for the full configuration reference.

Letting an AI Agent Do It

Deckyard is MCP-native, so a shell-capable coding agent can clone it, configure it with your keys, start it and wire itself in over MCP. The setup script takes flags for exactly that:

npm run setup -- --yes --ai=claude --ai-key=… --auth=off

Next Steps

Docker Deployment - the production path, with Compose
Configuration - customize your instance
Creating Presentations - start making presentations