Skip to content

Latest commit

 

History

2 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 

Repository files navigation

WorkBuddy Linux

中文 | English

Run Tencent's WorkBuddy AI office assistant (Windows build) natively on Linux.

How It Works

WorkBuddy is built on Electron. This project extracts app.asar and the resource files from the Windows installer, pairs them with the Linux Electron runtime, and ends up with a native Linux app — no Wine required.

Key techniques:

  • Extract the Electron app structure from the NSIS installer (.exe) using 7z
  • Replace the Windows native module (better-sqlite3) with a Linux build
  • Patch the load path for .node native modules inside the asar virtual filesystem

Prerequisites

# Debian/Ubuntu
sudo apt install p7zip-full rsync dpkg

# Node.js (v18+)
# nvm is recommended: https://github.com/nvm-sh/nvm
nvm install 22

Usage

1. Download the Windows installer

Download the Windows installer (.exe) from the WorkBuddy website.

2. Run the porting script

chmod +x port-workbuddy-to-linux.sh

# Basic usage
./port-workbuddy-to-linux.sh ~/Downloads/WorkBuddy-win32-x64-user-5.5.2.exe

# Custom install directory
./port-workbuddy-to-linux.sh ~/Downloads/WorkBuddy-win32-x64-user-5.5.2.exe /opt/workbuddy

The script performs the following 12 steps automatically:

  1. Extract the app files from the exe (7z unpacks the NSIS archive and the embedded 7z)
  2. Create the target directory
  3. Copy the resource files (excluding Windows binaries)
  4. Download the Linux build of Electron
  5. Unpack Electron
  6. Delete the Windows native modules (.node)
  7. Download the Node.js / Python runtimes
  8. Rebuild the better-sqlite3 native module (against the Electron ABI)
  9. Unpack app.asar
  10. Patch the asar load path in better-sqlite3
  11. Repack app.asar
  12. Generate the deb package

3. Install

# Install the deb (recommended)
sudo dpkg -i ~/workbuddy-linux/workbuddy-ai_5.5.2_amd64.deb

# Launch
workbuddy-ai

# Uninstall
sudo apt remove workbuddy-ai

Or skip the deb and run it directly:

~/workbuddy-linux/run-workbuddy.sh

Output

File Description
~/workbuddy-linux/workbuddy-ai_5.5.2_amd64.deb deb package
~/workbuddy-linux/run-workbuddy.sh Launcher script

After installation:

  • Command: workbuddy-ai
  • Desktop entry: search for "WorkBuddy AI" in the application menu
  • Data directory: ~/.config/WorkBuddy
  • URL schemes: wb-cover://, wb-extension://, workbuddy-ai://

Known Issues

  • You may need to log in again on first launch
  • If better-sqlite3 errors out, rebuild it manually:
    cd ~/workbuddy-linux/resources/app.asar.unpacked/node_modules/better-sqlite3
    npx @electron/rebuild --version 37.10.3 --module-dir .

Technical Details

Loading native modules from asar

Electron's asar virtual filesystem cannot load .node native modules directly (process.dlopen requires a real file path). asar pack --unpack "*.node" only excludes the files — it does not create redirect entries in the header.

The fix: modify better-sqlite3/lib/database.js to detect whether __dirname is inside an asar archive, and if so, build the .asar.unpacked path manually to load the native module.

Versions

Component Version
WorkBuddy 5.5.2
Electron 37.10.3
Node.js runtime 22.22.2

Further Reading

Two companion articles (in Chinese, by Y叔):

Article Content
我愿称之为「越狱版」WorkBuddy Running WorkBuddy under Bottles/Wine and bridging it to the host Linux toolchain (R / Python / git / make) via a file queue
没有枪没有炮我们自己造:真 Linux 版本 WorkBuddy Swapping the shell to build a true Linux build — the project in this repo

Invite Link

WorkBuddy currently offers free, time-limited access to several models, including DeepSeek 4.1, Hunyuan 3, and Hunyuan 4 Preview. Sign up here:

https://workbuddy.ai/invite?code=RSXQGMGW

Signing up through this link gives the inviter 100 credits and the new user 350 credits (credits can be exchanged for tokens).

License

This script is for personal study and research only. Copyright of WorkBuddy AI belongs to Tencent.

About

from exe to deb

Resources

Stars

4 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages