Running itRun at login
Run at login
Keep the worker and the scheduler running from login, on Linux with systemd and on macOS with launchd.
How to have Vendetta’s two long-running processes start by themselves, for whoever administers the computer. The workerWorker. The background process everything talks to, one per user. Start it with vendetta notify start. (vendetta notify) is what the chooser, the apps, and every plugin talk to. The schedulerScheduler. The process that runs schedules at their time. install.sh sets it to run at login. (vendetta schedule) runs schedules.
| Linux | macOS | |
|---|---|---|
| Scheduler | install.sh enables vendetta-schedule.service, a systemd user unit |
install.sh loads dev.vendetta.schedule |
| Worker | You enable a unit, below | install.sh loads dev.vendetta.notify |
Linux: the worker as a systemd user unit
Section titled “Linux: the worker as a systemd user unit”The checkout ships the unit. Copy it, stop a worker you started by hand so the unit can own it, and enable it:
mkdir -p ~/.config/systemd/user
cp integrations/vendetta-notify.service ~/.config/systemd/user/
vendetta notify stop
systemctl --user daemon-reload
systemctl --user enable --now vendetta-notify.serviceThe unit runs vendetta notify run with Restart=always and RestartSec=5, and KillMode=process, so agents and tmux servers outlive a worker restart.
Give it your PATH
Section titled “Give it your PATH”The worker finds provider executables (claude, codex, omp, …) through PATH, and a user unit does not have your shell’s. Add it with a drop-in:
systemctl --user edit vendetta-notify.service[Service]
Environment="PATH=/home/you/.local/bin:/usr/local/bin:/usr/bin"The scheduler’s unit gets the installer’s PATH when install.sh first writes it; an existing unit is kept as it is.
Stop and restart through systemd
Section titled “Stop and restart through systemd”systemctl --user restart vendetta-notify.service
systemctl --user stop vendetta-notify.service
journalctl --user -u vendetta-notify.servicevendetta notify stop alone is not enough here: the unit starts the worker again. Under a service manager the worker logs to the journal instead of worker.log.
Keep it running when you log out
Section titled “Keep it running when you log out”loginctl enable-lingerWith linger, your user manager starts at boot and stays after you log out, and the worker and scheduler with it. People’s Vendettas need this too.
Without systemd
Section titled “Without systemd”install.sh says so when it finds no systemd user manager. Start both yourself, from your session’s startup:
vendetta notify start
vendetta schedule startBoth are detached, and a repeated start is harmless.
macOS: launchd
Section titled “macOS: launchd”install.sh writes two launch agents to ~/Library/LaunchAgents and loads them:
| Label | Runs |
|---|---|
dev.vendetta.notify |
vendetta notify run |
dev.vendetta.schedule |
vendetta schedule run |
Each starts at login, is kept alive, and carries the PATH you ran install.sh with. Rerun install.sh after changing your PATH or installing a provider somewhere new.
To stop one until the next login or install:
launchctl bootout gui/$(id -u)/dev.vendetta.notifyProject machines after a reboot
Section titled “Project machines after a reboot”| Runtime | After a reboot |
|---|---|
| Docker | dockerd restarts the machines |
| Podman | Restarted only with linger and systemctl --user enable podman-restart.service, which vendetta machine project add enables for you |
Check it
Section titled “Check it”vendetta notify status
vendetta schedule status