Nova Filepond Laravel Package

Product Decisions This Supports

• Build vs. buy for admin upload UX. The field replaces a hand-rolled drag-and-drop widget, image previews, progress indicators, and client-side validation with a few lines of Nova configuration. For an internal admin surface this is almost always "buy": an uploader is table stakes, not a differentiator, and rebuilding one is weeks of undifferentiated work that produces no competitive advantage.
• Scoping a back-office efficiency line of work. FilePond enables a contained workstream — avatars, document attachments, media fields, single- and multi-image galleries — for ops and content teams. Frame it as a small enabler inside an existing roadmap, not a headline initiative, and resist letting it grow into a media platform.
• Rollout sequencing by risk. Adopt field by field: single-image uploads on a public disk with default paths are the low-risk pilot; multi-file galleries, custom load/store controllers, and tenancy-scoped authorization form a second wave that needs a deliberate regression pass. This gives a PM a natural phasing story without new scope.
• The dependency and bus-factor call. 55 stars, zero dependents, a single-maintainer pattern, and a last release of 2024-01-31 mean the real question is not "is FilePond's UX good?" (it is) but "are we prepared to own a fork if it stalls — especially across a Nova major version?" That answer should be explicit and named to an owner before the field becomes load-bearing on a critical workflow.
• Investment sizing and comparison. Typical cost is one small engineering ticket — Composer install, field configuration, disk and URL-resolution decisions — plus a staging QA pass covering upload, edit-screen preview, download, filename display, and prune behavior. That is the number to put next to the multi-week cost of a bespoke Nova upload field.
• Release and upgrade handling. v1.0.6 is a patch bugfix for filename return when a path is set in the Load Controller — maintenance hygiene, not a feature. It lets teams delete basename() display shims and directory-stripping workarounds, reducing code we own, but the removal must be a reviewed change: leftover shims can double-strip filenames.
• Data-model planning. The field stores relative paths, not URLs, so URL resolution must be owned deliberately — a model accessor/cast or a Nova display field. This is an architectural decision worth making before build starts rather than discovering in review, and it shapes how the field is used downstream in API responses and exports.

When to Consider This Package

Consider it when:

• The product already runs Laravel Nova as its internal admin surface, with Composer-managed dependencies and Laravel disk conventions for storage.
• Uploads are admin/internal, moderate volume, and single-tenant or cleanly scoped per tenant.
• The team wants FilePond's UX without owning frontend upload code, and no npm/mix build step is desirable — assets auto-register in Nova.
• Custom path or load-controller behavior is needed, and the team will validate filename handling in those configurations against v1.0.6 or later.
• A community package with a small maintainer base is acceptable, or the team has a named owner and is prepared to pin and fork.
• The requirement is presentation and plumbing for uploads — not asset transformation, rendition pipelines, or a DAM.

Look elsewhere when:

• The product is not on Nova — the value proposition collapses entirely.
• Uploads are public-facing or high-volume, where storage throughput, abuse controls, and rate limiting dominate the design.
• Requirements include enterprise DAM, asset transformation or rendition pipelines, or vendor SLAs and compliance guarantees.
• Multi-tenant or regulated data requires audited access control on the load/store endpoints beyond what the package ships.
• Compatibility against the target Nova/Laravel/PHP versions cannot be verified — a Nova major-version mismatch surfaces late as confusing frontend failures rather than a clean install error.
• The 55-star, single-maintainer profile is unacceptable for a workflow the business treats as critical, and no one has capacity to patch a fork.
• Chunked or large-video uploads are required in production and that path has not been exercised against your storage backend and web-server limits.

How to Pitch It (Stakeholders)

Executives. This is a low-cost buy-over-build accelerator for internal admin tooling. It buys a polished upload experience — drag-and-drop, previews, progress, async and chunked uploads, validation — instead of paying engineering to rebuild one. It is MIT-licensed: no procurement, no vendor contract, no new customer-facing surface, and no direct revenue impact. Frame it as a contained bet on back-office efficiency, not a platform commitment. Budget one small engineering ticket plus a QA pass, and be explicit that we may inherit maintenance if the community maintainer steps away. The current release is a maintenance bugfix, not a strategic shift. Exit cost is low — a Nova field is replaceable, and rollback is a pinned Composer version.

Engineering. A Laravel Nova field wrapping FilePond, installed via Composer with assets auto-registered — no build step. It writes relative paths into the column, so plan three things up front: an explicit disk (public plus storage:link, or S3 with temporaryUrl reads — omitting ->disk() inherits filesystems.default, usually local, and makes files browser-unreachable), authorization and signed-URL behavior on the Load Controller, and the scheduled prune of temporary uploads. The published config (default disk, temp path, route names, route middleware) is the sanctioned override surface; beyond it, subclass the field and override load()/store() rather than editing vendor controllers — there are no events to hook. Validation is client-side only and advisory (maxFileSize is in bytes, not KB), so duplicate the limits in server-side rules. v1.0.6 corrects the filename returned by the Load Controller when a custom path is set, so previews show photo.jpg rather than avatars/photo.jpg; no breaking changes are indicated, but remove any now-redundant basename() shims deliberately and re-test default and custom-path flows. Pin the version, confirm the Nova/Laravel/PHP support matrix before committing, and monitor upload/load error rates for a release cycle — broken previews (401/403 on load) and non-public disks are the two recurring failure modes.