**System details:** Omarchy `quattro` @ 7488eade (post-4.0.0). Found by an automated QA pass. Proven in source only — not reproduced, since it needs a Surface device with a discrete NVIDIA GPU.

`install/hardware/fix-surface-keyboard.sh:20` writes `MODULES=(...)` rather than `MODULES+=(...)`:

```sh
echo "MODULES=(${pinctrl_module} surface_aggregator surface_aggregator_registry surface_aggregator_hub surface_hid_core surface_hid surface_kbd intel_lpss_pci 8250_dw)" > \
  /etc/mkinitcpio.conf.d/surface_device_modules.conf
```

mkinitcpio sources drop-ins in lexical order, and `surface_device_modules.conf` sorts after `nvidia.conf`, which `install/hardware/nvidia.sh:31` writes with `MODULES+=(nvidia nvidia_modeset nvidia_uvm nvidia_drm)`. So on a Surface Book 2/3 or Surface Laptop Studio — all of which ship a discrete NVIDIA GPU — the nvidia modules are dropped from the initramfs and early KMS never happens, which is the usual cause of the boot-time black frame or mode flip on those machines.

The tree already knows this file resets `MODULES`. `etc/mkinitcpio.conf.d/omarchy_hooks.conf:12-16` says so explicitly:

```
# This reads MODULES midway through mkinitcpio's drop-in sourcing, so a
# later-sorting drop-in that resets MODULES outright — surface_device_modules.conf
# does — would strip nvidia_drm after kms was already dropped. Every machine
# Omarchy writes such a file for carries an Intel iGPU, which keeps kms here
# through the scan below; keep it that way.
```

That mitigation preserves the `kms` hook, which keeps the iGPU working — but it does not put the nvidia modules back, so NVIDIA early KMS is still lost.

Same class as #6876 (`omarchy_hooks.conf` replaces `HOOKS` instead of extending it), different variable and different file. `install/hardware/apple/fix-spi-keyboard.sh:9,12` has the same `MODULES=` shape; there it is currently harmless because the SPI MacBook models it matches have no discrete GPU and no other Omarchy drop-in to clobber.

---

**Correction, after an independent review of this report:**

Narrower than filed. It requires a Surface that passes detection *and* a successful `pinctrl_` autodetect. The visible black-frame symptom is not established from source, since the Intel iGPU keeps the `kms` hook -- the provable consequence is that the NVIDIA modules are dropped from subsequently generated initramfs images. The MacBook aside is also wrong: the SPI regex matches `MacBookPro13,3`, which does have a discrete AMD GPU, so that file is merely non-conflicting today rather than harmless by construction.