# notifyd

A freedesktop notification daemon for this desktop, replacing dunst.

The daemon owns `org.freedesktop.Notifications` and holds the state. It
publishes the live queue and the history ring as JSON under
`$XDG_RUNTIME_DIR/notifyd/` for a quickshell renderer to draw, and `notifyctl`
is the only thing that talks back over D-Bus.

    go build ./...

The design and the exact file contract are in the quickshell repo, at
`docs/superpowers/specs/2026-09-15-notification-daemon-design.md`.

## Install and handover

Build and install the two binaries and the script, beside the statusctl CLI:

    go build -o ~/bin/notifyd ./cmd/notifyd
    go build -o ~/bin/notifyctl ./cmd/notifyctl
    install -m 755 scripts/notify-snooze.sh ~/bin/notify-snooze.sh

dunst is not removed until this proves itself. To switch:

1. Stop dunst (`pkill -x dunst` or its service) so the bus name is free.
2. Add `hl.exec_cmd("notifyd")` to `~/.config/hypr/sections/autostart.lua`,
   beside the quickshell lines.
3. Change `rofipass`'s one `dunstctl close-all` to `notifyctl close-all`.
4. Start `notifyd` and send a test notification.

The renderer that draws the balloons is a separate plan; until it ships the
queue is visible through `notifyctl list`.

## Configuration

`~/.config/notifyd/config.json` (`$XDG_CONFIG_HOME` is honoured). Every key
is optional and the defaults are the behaviour without the file:

    {
      "timeouts": { "low": 10, "normal": 10, "critical": 0 },
      "liveCap": 20,
      "historyCap": 20
    }

Timeouts are seconds, 0 is never, and apply only when a client sends
`expire_timeout -1`. Caps clamp to 1..100; a lowered cap applies at the next
notification. A broken file is logged and read as the defaults.

The daemon reads it at start and on `notifyctl reload`. The quickshell
settings page writes it; the same file carries the balloon shell's keys
(`balloons`, `dndPassCritical`, `muted`), which the daemon ignores.

## License

GPLv2 only. See [LICENSE](LICENSE).

## Development Approach

This project is developed using AI-assisted tools. Code is generated with the help of AI based on human-provided specifications, design decisions, and iterative feedback.

All contributions are reviewed, tested, and curated by the maintainer before being included in the codebase. AI is used as a productivity and exploration tool, while human oversight remains central to all decisions.

The goal is to combine the flexibility of AI-assisted development with standard open-source practices such as transparency, review, and accountability.