Statamic Site Weather Laravel Package

Product Decisions This Supports

• Status surface for internal dashboards. The package gives an existing Laravel dashboard a single glanceable "weather" tile that rolls up several independent health signals, instead of a table of raw checks that no one reads.
• Build vs. buy for status sources. The WeatherContributor contract plus the site-weather.contributors container tag means each new data source (uptime provider, CI pipeline, incident feed, deploy status) is registered as a contributor rather than forked into the widget. That is the reusable seam; the Blade tile is commodity UI.
• A policy decision about honesty, encoded. Worst-wins aggregation with unknown as a first-class state means the tile can say "we don't know" rather than defaulting to healthy. "An empty tile never says sunny" is a defensible product stance: absence of data is not evidence of health. Teams should confirm this matches their on-call expectations before adopting.
• Accessibility without a separate audit cycle. The widget ships with a distinct icon shape per state (not colour-only encoding), a word for every state, and a one-sentence screen-reader summary, and was axe-audited on the real dashboard in both colour schemes with no violations. That removes a recurring compliance task from the roadmap for this surface.
• Roadmap seed for a broader status page. Reading already carries headline, url, and computed_at, so drill-down links and freshness display are available without changing contributor contracts. Public status pages, history, and alerting are not in scope but are not blocked by the data model.
• Release-readiness framing. "Is it safe to deploy right now?" is a natural composition of deploy, CI, and incident contributors into one worst-wins tile — a decision-support use case rather than a monitoring one.

When to Consider This Package

Adopt when:

• You already have a Laravel dashboard or widget surface (custom Blade, Nova, Filament, etc.) and a reason to add a status tile.
• You have two or more independent health signals that genuinely should roll up into one worst-wins verdict.
• Accessible, colour-blind-safe status encoding matters and you would otherwise rebuild the icon/word/screen-reader treatment yourself.
• Multiple teams own different signals and you want them to contribute without touching the widget.

Look elsewhere when:

• You need time-series storage, alerting, paging, or SLO/error-budget maths. This is a display contract; computed_at is a timestamp, not a metric.
• You need per-contributor history or trend charts — nothing in this release provides it.
• You need a full status-page product (subscriber notifications, incident timelines, public SLAs).
• You are rendering through a non-Blade front end (React/Vue SPA); the widget is described as Blade and static.
• You require a proven long-term API stability commitment. This is v1.0.0 — the first release, with no upgrade path exercised. If you pin to it, pin exactly and expect early-adopter churn.

Validate before shipping: the notes do not describe caching, timeouts, or async evaluation for contributors, and the widget appears to render computed readings. Confirm contributor latency and failure behaviour against your dashboard's page budget, and decide what a slow or throwing contributor should render (presumably unknown).

How to Pitch It (Stakeholders)

For executives. This is a small, contained package that turns several operational signals into one trustworthy red/amber/green tile on an internal dashboard. Its distinguishing property is honesty: it never reports green when a source is unknown, so the tile cannot lull anyone into thinking everything is fine when a check is simply missing. It has also been accessibility-audited — distinct icon shapes, a word per state, and a screen-reader summary, verified with axe in both colour schemes with no violations — which retires a recurring compliance concern for this surface. Risk is low and scope is narrow: it is UI, not monitoring, and it will not page anyone.

For engineering. Integration is one contract and one container tag: implement WeatherContributor, register it under site-weather.contributors, return a Reading of state, headline, url, and computed_at. Each owning team ships its own contributor with no changes to the widget. Aggregation is worst-wins with unknown as a first-class state — treat that as fixed semantics rather than configuring per-contributor overrides, because the whole point is that the tile's verdict is predictable. The url field keeps the tile drillable, so contributors should supply the most useful destination for debugging.

For the deciding conversation. Buy the contract and the aggregation semantics; buy the widget only if you don't already have one. The thing worth owning is the shared definition of "what does green mean," and this package makes that definition explicit and conservative. Budget for verifying contributor latency and caching on a high-traffic dashboard, and treat v1.0.0 as an early-adopter dependency until a second release confirms the API.