Nova Dark Theme Laravel Package

Product Decisions This Supports

• Whether the admin surface deserves UX polish budget at all. Nova is where internal operators spend long shifts; a dark theme is a cheap, visible ergonomics win for internal users. This package lets you test that hypothesis with a one-ticket spike instead of a sprint, and gives you a concrete demo to put in front of stakeholders.
• Build vs. buy on a commodity feature. A dark mode with a persisted toggle is well-understood, undifferentiated scope. v1.0 delivers the whole of it — a Vue toggle component, an on/off preference, and Artisan commands that scaffold the control into Nova templates. That is the buy case: one Composer require and an Artisan command instead of a ticket. It is also the build case: the same thing is a CSS-variable sheet plus localStorage and a stored preference, a few dozen lines you fully own.
• Roadmap triage on a stale dependency. Last tagged release February 2019, 71 stars, zero dependents. That is not a package tracking Nova's release cadence — it is one frozen at an old Nova major. This supports a defer/reject for any current Nova installation and a timeboxed spike only for a legacy panel on the era it was built for. It is not a candidate for the roadmap; it is a candidate for an evaluation ticket.
• Scope discipline on theming. Adopting this is an implicit "no" to design tokens, per-tenant branding, and high-contrast variants. A fine decision if made deliberately; a bad one if a stakeholder later assumes the toggle generalizes into a theming system. Say that out loud in the kickoff.
• Accessibility and ergonomics commitments. A user-controllable theme is a tangible line item in a UX or accessibility review (light sensitivity, low-light operation). Note that the switch's own accessibility — keyboard focus, role="switch"/aria-pressed, label text, prefers-color-scheme fallback — is unverified and becomes your team's responsibility, not the maintainer's.
• Dependency governance precedent. A clean test case for how the org treats abandoned community packages: pin exactly, capture the scaffold diff, flag-gate, timebox, record an ADR. Whatever rule you apply here should be the rule for the next one — that consistency is worth more than this package.

When to Consider This Package

Consider it when most of the following hold:

• You run Laravel Nova, specifically a major from the package's era. The Artisan commands that wire a toggle into Nova templates are the only genuinely Nova-specific value here. Verify on a throwaway branch rather than trusting the README.
• You want explicit user control, not an OS-level prefers-color-scheme media query. v1.0's contribution is a deliberate on/off switch that persists — that is the product decision it encodes, and it's a defensible one for shared operator terminals.
• Your front end is Vue at the major Nova bundles. The toggle ships as a Vue component; on Livewire, Inertia+React, or plain Blade you are writing the glue and the leverage collapses.
• The blast radius is acceptable. It should be: a theme class plus a stored preference. If the scaffolding touches published Nova views, capture the diff first — that diff is both your rollback artifact and your conflict baseline for the next Nova upgrade.
• You can pin an exact version and own the code. At this cadence, treat it as a vendored snippet with a Composer wrapper, not a maintained dependency.

Look elsewhere — or write it yourself — when:

• You are on a current Nova major (4.x/5.x). A 2019 tag almost certainly predates the Vue and template changes Nova has shipped since; the compatibility claim cannot be trusted without a fresh test, and Nova is closed-source so you cannot patch around incompatibilities.
• Your stack is not Nova + Vue. CSS variables plus localStorage is a few dozen lines, has no upstream, and reverses trivially. At zero dependents, the ecosystem has effectively already voted.
• You need a theming system — tokens, per-tenant branding, high-contrast modes. This is a toggle, not a design system.
• You operate strict dependency governance (security review, license audit, maintenance SLA). MIT clears licensing; maintenance clears neither of the others. There is no data here on test coverage, issue responsiveness, or uninstall path.
• You run strict CSP, no-JS, or SSR admin surfaces, where a Vue toggle and flash-of-wrong-theme need deliberate handling the release notes do not address.

How to Pitch It (Stakeholders)

For executives (one line): A tiny, free, MIT-licensed Nova theme that gives operators a dark-mode switch — worth a one-day spike on the admin panel, and worth skipping entirely if our Nova version has moved past it or a real theming effort is already planned.

For engineering leadership: Narrowly scoped presentation-layer package. v1.0 ships a persisted dark/light preference, a Vue toggle component, and Artisan commands that scaffold the control into Nova templates. It does not touch routing, persistence schemas, queues, auth, or business logic, so architectural coupling is low and review is quick. Two caveats must be stated plainly, because they change the recommendation:

1. It is stale. Last release February 2019, zero dependents, 71 stars. That is not a maintained dependency — it is frozen at an old Nova major. Do not build a plan on assumed compatibility with the Nova we actually run; test it first.
2. The Artisan commands edit files. Fast to adopt, awkward to audit, manual to reverse. Capture the diff of every touched file before merge, confirm a documented uninstall exists, and confirm whether a preference column is introduced — and whether that migration is nullable and reversible.

Recommendation: Timebox a one-ticket evaluation on a single non-production Nova panel, pinned to an exact version and flag-gated for one release cycle. If it installs cleanly and the diff is contained, pilot on the internal admin panel. If it fights the Nova version or the diff sprawls into published views, stop and spend the half-day on a small internal CSS-variable override instead — the surface area is small enough that in-house ownership is cheaper than carrying an abandoned dependency, and it removes the annual Nova-upgrade breakage risk entirely. Do not extend this to anything customer-facing.