Filament Undo Toast Laravel Package

Product Decisions This Supports

• Invest in destructive-action safety, not just confirmation dialogs. This release makes undo a first-class safety net for both soft deletes and hard deletes — including restoring related rows (BelongsToMany, HasMany, HasOne, MorphMany, MorphOne) in the same transaction. That turns "are you sure?" friction into recoverable action, which is a product decision about how much we trust operators versus how much we gate them.
• Build vs. buy for admin UX polish. The toast behaviour here (stacking, progress line, dark mode, High Contrast, touch fan-out, mod+z, failure states announced as alerts) is the kind of detail teams routinely under-scope and then ship badly in-house. Adopting this is a cheap way to buy a polished interaction instead of speccing one.
• Global rollout readiness. 23 locales with correct plural forms (Russian, Ukrainian, Polish, Czech included) and Filament's own terminology reused per language means supporting a multi-region ops team isn't a follow-on localisation project.
• Accessibility and compliance commitments. Windows High Contrast support, alert-role announcements on failed undo, and keyboard operability while focus is inside a checkbox/radio/button are concrete evidence for an accessibility roadmap item rather than a promise.
• Reducing duplicate/confusing UI. The default now suppresses Filament's own success notification when an undo entry exists, with per-action opt-out (->replaceSuccessNotification(false) or a closure). That's a decision about notification noise and operator attention, not just styling.
• Roadmap sequencing. The addition of a Testbench workbench app, a Playwright end-to-end suite in CI, SECURITY.md, Dependabot with cooldown and SHA-pinned Actions signals a maintainer investing in longevity — relevant when deciding whether to depend on it for a multi-quarter roadmap.

When to Consider This Package

Consider it when:

• The product surface is Filament admin panels and destructive actions are routine (ops, support, content moderation, back-office tooling).
• Users are internal operators who act fast and occasionally misclick; undo measurably beats confirmation modals for their workflow.
• You need hard-delete recovery with related records, and you want it transactionally consistent rather than hand-rolled.
• You're shipping to multiple locales, or plan to, and don't want to own translation and pluralisation for a UI affordance.
• Accessibility is a stated requirement or procurement gate.
• You want a small, auditable dependency rather than building and maintaining undo semantics, toast state management and keyboard handling yourselves.

Look elsewhere when:

• You're not on Filament — this is tightly coupled to Filament's notification and action lifecycle; there is no meaningful reuse in a custom UI or headless/API context.
• You need durable audit trails or regulatory retention. An undo toast is ephemeral by design; it is not a substitute for soft deletes, event sourcing, or an audit log. If deletion must be provably reversible for days, model that in the data layer.
• You need bulk or multi-user undo semantics — undo is oriented around a single recent action by the acting user, not collaborative reversal.
• Hard deletes are rare or already gated behind approval workflows; the added surface buys little.
• Your organisation requires vendor support contracts or a formal security review process the package can't satisfy, however well the maintainer has set up Dependabot and SECURITY.md.

How to Pitch It (Stakeholders)

For executives: A small, focused add-on for our admin tooling that makes deleting things recoverable instead of merely confirmed. It removes a class of "I deleted the wrong record" incidents and the support tickets and rework that follow, including restoring linked records like pivot and child rows, not just the one record. It ships translated into 23 languages with correct grammar for our non-English teams, meets accessibility expectations including Windows High Contrast, and is backed by an automated end-to-end test suite and a published security policy. The cost is adoption and a small ongoing dependency; the alternative is building and maintaining equivalent interaction, localisation and accessibility work ourselves.

For engineering: One trait or a per-action configuration, opt-out available via ->replaceSuccessNotification(false) or a closure so we don't lose bespoke notifications. Relation restoration runs in the same transaction and only inserts missing rows, so it's safe to call on partially-restored data. A Testbench workbench and Playwright suite run in CI, so we can validate behaviour before rollout rather than eyeballing toasts. Language files are keyed and covered by a test that checks placeholders and plural rendering per locale.

For design/UX: The interaction has been through a design review pass — progress line inside the rounded corners, stable stack height and tint, aligned Undo affordances, persistent failure states, touch-friendly stack fan-out, will-change scoped to motion only. We get a considered pattern rather than a component we style from scratch, and it keeps Filament's visual language.