## What happened

On a laptop + external monitor setup where the internal panel (`eDP-1`) has
been soft-disabled via `omarchy-hyprland-monitor-internal off` (external-only
mode), physically unplugging the external monitor does **not** re-enable the
internal panel. The screen goes black instead of falling back to the laptop
display, even though `omarchy-hyprland-monitor-watch` is running and the
`monitorremoved` event fires correctly.

## Root cause

When the external monitor is unplugged while `eDP-1` is config-disabled,
Hyprland/aquamarine has no remaining real output and creates a synthetic
`FALLBACK` headless monitor so it isn't left with zero monitors.

`omarchy-hyprland-monitor-external-active` (in
`omarchy-hyprland-monitor-clamshell`) checks:

```bash
hyprctl monitors all -j | jq -e '.[] | select(.name | test("^(eDP|LVDS|DSI)-") | not) | select(.disabled == false)'
```

This matches `FALLBACK` too (it isn't named `eDP*`/`LVDS*`/`DSI*` and isn't
disabled), so the function reports "an external monitor is active" even
though the real one was just removed. `omarchy-hyprland-monitor-internal`'s
`recover()` then short-circuits on that check and never removes the
`internal-monitor-disable` toggle flag, leaving `eDP-1` disabled forever
(until the user manually intervenes).

## Steps to reproduce

1. Laptop with an internal panel + an external monitor connected.
2. `omarchy-hyprland-monitor-internal off` (disable internal, external-only).
3. Physically unplug the external monitor's cable.
4. Expected: internal panel re-enables automatically.
5. Actual: screen stays black. `hyprctl monitors all -j` shows `eDP-1` still
   `"disabled": true` and a `FALLBACK` monitor in its place.

Confirmed live via `socat -U - UNIX-CONNECT:$SOCKET` on `.socket2.sock`: the
correct `monitorremoved>>HDMI-A-1` / `monitorremovedv2>>...` events do fire,
but `omarchy-hyprland-monitor-external-active` still exits 0 immediately
after, because of the `FALLBACK` monitor.

## Evidence

Live capture of Hyprland's socket2 IPC events at the moment of unplugging,
with the state of `omarchy-hyprland-monitor-external-active` and the toggle
flag checked immediately after each event:

```
22:42:26.577910760 EVENT: monitorremoved>>HDMI-A-1
  external-active exit=0
  toggle-flag-present=yes
    eDP-1 disabled=True
    FALLBACK disabled=False
22:42:26.615480742 EVENT: monitorremovedv2>>2,HDMI-A-1,LG Electronics LG ULTRAGEAR+ 512NTQDGB572
  external-active exit=0
  toggle-flag-present=yes
    eDP-1 disabled=True
    FALLBACK disabled=False
22:42:26.650419651 EVENT: monitoradded>>FALLBACK
  external-active exit=0
  toggle-flag-present=yes
    eDP-1 disabled=True
    FALLBACK disabled=False
22:42:26.689077230 EVENT: monitoraddedv2>>1,FALLBACK,
  external-active exit=0
  toggle-flag-present=yes
    eDP-1 disabled=True
    FALLBACK disabled=False
```

`external-active exit=0` (true) right after `HDMI-A-1` is removed and only
`FALLBACK` remains confirms the check is fooled by the synthetic monitor,
which is why `internal-monitor-disable`'s toggle flag (`toggle-flag-present=yes`)
never gets cleared.

## Expected behavior

`omarchy-hyprland-monitor-external-active` (and any other check relying on
"is a real external monitor connected") should exclude Hyprland's synthetic
`FALLBACK` output, e.g.:

```bash
hyprctl monitors all -j | jq -e '.[] | select(.name != "FALLBACK") | select(.name | test("^(eDP|LVDS|DSI)-") | not) | select(.disabled == false)'
```

## System info

- Omarchy: 4.0.2-1
- Hyprland: 0.56.2 (commit efb50993780079460b0cbed1363e2166a2de1d9f)
- CPU: AMD Ryzen 7 6800H
- GPU: NVIDIA GeForce RTX 3050 Ti Mobile (GA107M) + AMD Radeon 680M (Rembrandt) — hybrid graphics laptop
- Monitors: internal `eDP-1` (1920x1200@165) + external `HDMI-A-1`, LG UltraGear 4K (3840x2160@144)

## Workaround

Until this is fixed upstream, a small user-level watcher script listening for
the specific external monitor's `monitorremoved` event, excluding `FALLBACK`
from the active-check, and clearing the toggle flag + `hyprctl reload` when no
real external remains. Save as e.g. `~/.config/hypr/scripts/external-monitor-recovery-fix.sh`,
set `EXTERNAL` to your external monitor's connector name (`hyprctl monitors`),
`chmod +x` it, and start it with `o.launch_on_start(...)` in
`~/.config/hypr/autostart.lua`:

```bash
#!/bin/bash
# Fixes a bug in Omarchy's own laptop-panel recovery: when eDP-1 is soft-disabled
# (via `omarchy-hyprland-monitor-internal off`) and the external monitor is
# unplugged, Hyprland creates a synthetic "FALLBACK" headless output so it's
# never left with zero monitors. Omarchy's `omarchy-hyprland-monitor-external-active`
# check doesn't exclude FALLBACK, so it thinks an external monitor is still
# active and never re-enables eDP-1 -- the screen stays on the invisible
# FALLBACK output. This script watches for the specific external monitor being
# removed and, if no *real* external monitor remains, re-enables eDP-1 itself.

EXTERNAL="HDMI-A-1"  # <- set to your external monitor's connector name
LOCK="${XDG_RUNTIME_DIR:-/tmp}/omarchy-external-monitor-fix.lock"
TOGGLE_FLAG="$HOME/.local/state/omarchy/toggles/hypr/internal-monitor-disable.lua"
SOCKET="$XDG_RUNTIME_DIR/hypr/$HYPRLAND_INSTANCE_SIGNATURE/.socket2.sock"

exec 9>"$LOCK"
flock -n 9 || exit 0

real_external_active() {
  hyprctl monitors all -j | jq -e \
    '.[] | select(.name != "FALLBACK") | select(.name | test("^(eDP|LVDS|DSI)-") | not) | select(.disabled == false)' \
    >/dev/null 2>&1
}

recover_if_stuck() {
  [[ -f $TOGGLE_FLAG ]] || return 0
  real_external_active && return 0
  rm -f "$TOGGLE_FLAG"
  hyprctl reload >/dev/null 2>&1
  hyprctl dispatch 'hl.dsp.dpms({ action = "enable" })' >/dev/null 2>&1
}

while read -r event; do
  case "$event" in
    monitorremoved">>"$EXTERNAL | monitorremovedv2">>"*,$EXTERNAL,*)
      # Give Hyprland a moment to settle on FALLBACK before checking.
      sleep 0.3
      recover_if_stuck
      ;;
  esac
done < <(socat -U - "UNIX-CONNECT:$SOCKET")
```