Getting started

There are three ways to run Deskforce: the desktop app, from the source with Node.js, or with Docker.

Desktop app

Installers are on the Releases page. Nothing else needs to be installed; enter your API keys in the app's settings.

System	File
macOS, Apple Silicon (M1 and later)	Deskforce-<version>-arm64.dmg
macOS, Intel	Deskforce-<version>-x64.dmg
Windows 10 and 11	Deskforce-Setup-<version>.exe
Linux, any distribution	Deskforce-<version>-<arch>.AppImage
Debian, Ubuntu and derivatives	Deskforce-<version>-<arch>.deb

• macOS: the app is signed and notarised by Apple. Open the .dmg, drag Deskforce to Applications, and start it like any other app.
• Windows: the installer is not signed with a paid certificate yet, so Windows asks before it runs it the first time. When "Windows protected your PC" appears, choose More info → Run anyway.

On Linux (installers from version 0.4.0; for earlier versions use Docker):

• AppImage: make it executable and start it: chmod +x Deskforce-*.AppImage && ./Deskforce-*.AppImage. On Ubuntu 22.04 and later it needs FUSE 2: sudo apt install libfuse2 (libfuse2t64 on 24.04).
• .deb: sudo apt install ./Deskforce-*.deb, then start Deskforce from your applications menu.

More on the app, building it yourself, signing and updates: Desktop app.

From the source

1. Prerequisites

• Node.js 22 or newer (nvm use picks it up from .nvmrc)
• At least one AI provider:
  • an API key for OpenAI, Anthropic, Google Gemini or xAI (Grok), or
  • Ollama running locally with the llama3 model (ollama pull llama3) — set OLLAMA_MODEL to use a different one

2. Install

git clone https://github.com/husseinzayat3/Deskforce.git
cd Deskforce
npm install

3. Configure

Copy the example environment file and add the keys for the providers you plan to use:

cp .env.example .env.local

.env.local is ignored by git. A provider without a key returns mock responses, so you can explore the interface before adding one. Every setting is described in .env.example.

You can also enter keys, models and the Ollama address in the app itself: the gear button opens the settings. Values saved there are kept in the office database and take precedence over .env.local.

Keys are stored unencrypted. API keys and Slack tokens saved in the settings are written as plain text to the office database (workspace/office_state.db, or office_state.db in the desktop app's data folder), just as they are plain text in .env.local. Anyone who can read that file can read the keys, so keep it out of backups and folders you share, and never commit or send it.

4. Run

npm run dev

Open http://localhost:3000, hire a few employees, and assign a task to the most senior one.

With Docker

docker run -p 127.0.0.1:3000:3000 -v $(pwd)/workspace:/app/workspace --env-file .env.local ghcr.io/husseinzayat3/deskforce

See Docker for building the image yourself and using Ollama from the container.

Next

• Features — everything the office can do
• How it works — the engine, the tools agents use, and the technology
• Slack — talk to the office from Slack
• Security model — read this before exposing the office to anything but your own computer