The Razer peripherals' tray on the workstations, as a module (novox/hq ADR 0208). It requires x11-display, so it is assigned only where a display server is held on the same machine. The driver and the daemon it drives are the openrazer module's.

Owns

what	where
the tray's one start	a contribution to `node-display-session` (ADR 0212): `exec --no-startup-id polychromatic-tray-applet`, placed by the `i3` module among the other modules' lines

Nothing else. It declares no package, holds no seat and writes no file.

The application is kept as found. polychromatic (0.9.8) is not in the official repositories: on both workstations it is a foreign (AUR) package, installed explicitly. The host installs from the official repositories only, so the module cannot declare it. ADR 0205's pinned archive does not fit either: it is a Qt application with a helper, a tray and a controller, not a set of plain files. Like snapd and the laptop's triggerhappy, it waits for the mesh's package repository (research 027 question 1, option P2). Until then a fresh workstation installs it by hand, and polychromatic_check says when it is missing.
Its dependencies are the openrazer module's (python-openrazer, the daemon, the driver). The tray brings them when installed by hand; the module does not declare them twice.
The application's settings stay the operator's. ~/.config/polychromatic/ (preferences, presets, effects, device states) is the application's own, rewritten by it. The tools read only the tray's two keys of preferences.json.

How it starts: the module's window-manager line, and nothing else

The tray had two starts on both workstations:

the window-manager line exec --no-startup-id polychromatic-tray-applet, in the i3 module's section Until their modules carry them;
the package's XDG autostart entry polychromatic-autostart.desktop, which dex runs at login. It runs polychromatic-helper --autostart, which waits for the daemon, resumes each device's software effect, and starts the tray too while the application's setting Start the tray applet when I log on is ticked. It is ticked on both.

The tray takes a lock file at its start and stops an earlier instance, so two starts end as one tray, in a race. On the laptop both ran at the login of 2026-10-04 16:26, and one tray was left; which start it came from is not recorded anywhere. On the desktop only the window-manager line ran (that session began before dex was installed).

This module takes over the window-manager line as its contribution, and the i3 module drops it in the same change. That line is the tray's one start:

it starts the tray with the session, once (exec, so a reload of i3 starts nothing);
it is the mesh's, so it is in the composed configuration only where the module is assigned.

The helper stays, for its other work: the effects it resumes at login. The package's entry is not the module's to hide, and hiding it would need a file in the account's autostart directory. So the operator unticks the tray setting once (migration, below), and polychromatic_check names the helper's tray start as a second start for as long as the setting is ticked. The module never writes the setting: the application rewrites its preferences file itself.

Tools

They are served by the node's runtime as the operator account (ADR 0175).

tool	does
`polychromatic_status` (r)	whether the tray runs: pid, since, and the scope or unit it runs in the installed version, and that it is from outside the official repositories what starts it: the window-manager lines, and the helper while the setting is on the helper's entry, and the tray setting with its delay whether the openrazer daemon answers, and how many devices it has (asked with `--auto-start=no`: never starts it)
`polychromatic_restart` (a)	asks the tray to end (SIGTERM), forces it after 5 s, and starts `polychromatic-tray-applet` in the operator's session as a transient user unit `mesh-polychromatic-tray`, so it outlives the tools runtime. Refused plainly when nobody is logged in to the desktop
`polychromatic_check` (r)	the package is installed exactly one start: one window-manager line; the helper does not start the tray too; no autostart entry of the account for it one tray runs in a desktop session the openrazer daemon answers Each finding says what to do

The kernel keeps 15 characters of a command name, and polychromatic-tray-applet and this bundle's polychromatic-tools share them. The tools match the tray by its full program name, so they never count or end themselves.

What changes when it is assigned

	laptop	desktop
package	none: `polychromatic` 0.9.8, explicit, foreign	the same
`~/.config/i3/config`	the tray's line moves from the `i3` module's section to this module's contribution: the same line, and i3's reload starts nothing	the same
tray	none: one runs, left by the two starts' race	none: one runs, from the predecessor's window-manager line (a child of i3, since the session of 2026-10-04 16:00, before `dex` ran)

Migration (ADR 0182)

On each workstation, once: in Polychromatic, untick Preferences → Tray → Start the tray applet when I log on. From the next login the module's line is the tray's one start, and polychromatic_check answers ok once the daemon answers too (the openrazer module's migration).

Leaves as found

~/.config/polychromatic/: preferences, presets, custom effects, device states.
/etc/xdg/autostart/polychromatic-autostart.desktop, the package's.
The package itself, installed by hand.

Relies on

The openrazer module for the daemon and the driver. The tray without them shows no devices. polychromatic_check reports it.
The i3 module to place the contribution (ADR 0212 §4: the contribution depends on node-display-session), and to run dex for the helper.
A display server on the same machine (x11-display, ADR 0208 §3).