Nova Aegis Laravel Package

Product Decisions This Supports

• Roadmap: ship trust-sensitive features without owning the plumbing. The v2.1.0 notes show a package that already owns the hard parts of a scan → check → audit pipeline: a target allowlist with a deliberate 422 contract (TargetNotListed), deterministic check execution, and an audit log. That maps to "build vs. buy" calls on compliance/verification surfaces — buy the engine, build the product around it.
• Feature: auditable, time-stamped evidence. Checks and the audit now read time from an injected PSR clock. This is the difference between "we ran a scan" and "we can prove, reproducibly, what was true at time T." It unlocks customer-facing audit exports, SLA/schedule reporting, and test suites that assert on time without sleeping — decisions that were previously blocked or deferred.
• Feature: safe, API-addressable workflows. Input is read through form requests, so the API surface is validated and documented at the boundary rather than deep in the domain. This reduces the cost of exposing scanning to third parties, CI pipelines, or a separate frontend.
• Correctness bet: no phantom state. SettingsSaved now fires after the transaction commits, so listeners never consume values that later roll back. This directly affects any decision to build integrations on top of that event (webhooks, cache invalidation, provisioning, billing triggers) — those integrations can now be treated as reliable rather than eventually-ish consistent.
• Release policy signal: v2.1.0 is explicitly a non-breaking internal refactor — "no change for applications." That is a maturity datapoint worth recording: the maintainers are willing to restructure internals without passing migration cost to consumers, which lowers the perceived risk of pinning to a minor-version upgrade path.

When to Consider This Package

Adopt when most of the following hold:

• You need an allowlist-gated scanning/verification capability and are comfortable with an explicit 422 + TargetNotListed rejection contract rather than silent skipping.
• Your use case requires deterministic, clock-injectable timestamps — scheduled audits, reproducible test assertions, regulated evidence trails, or customer-facing reports.
• You want the settings/config write path to be transactional and event-safe, and you intend to attach listeners (SettingsSaved) to it.
• You are on a maintained Laravel version and want form-request-level validation on the HTTP surface without writing it yourself.
• You value a maintainer who refactors internals behind a stable public contract.

Look elsewhere or wait when:

• You need to scan arbitrary, user-supplied targets. The allowlist is a hard gate here, not a config toggle; if dynamic target registration isn't supported, this is a poor fit.
• Your audit requirements demand an external, tamper-evident store (WORM, append-only ledger) — an in-app audit table may not satisfy an auditor regardless of clock injection.
• You need a documented public API/extension contract for checks. The release notes describe internal restructuring; confirm which extension points are considered stable before building a plugin ecosystem on them.
• You carry a large existing listener surface that assumes synchronous, pre-commit settings semantics. The after-commit dispatch is the right behavior, but audit listeners that ran inside the transaction should be re-verified.

How to Pitch It (Stakeholders)

To executives: This dependency gives us a scan → verify → audit capability with reproducible, time-stamped evidence and a validated API surface, without us building or maintaining that core. v2.1.0 is a maintenance release — no breaking changes, no migration work, no application behavior changes — which is the signal we want from a vendor: they improve internals on their own dime. Our risk is concentrated in one policy decision (the target allowlist), which is a deliberate security posture, not a limitation we can configure around.

To engineering: Upgrade is safe and should be treated as routine; nothing in the application contract moved. Three things worth knowing: (1) unlisted targets are still rejected with the same 422/TargetNotListed response — the exception class is now part of the observable contract, so keep it in your assertions; (2) checks and audit read time from an injected PSR clock, so freeze the clock in tests instead of tolerating time-based flake; (3) SettingsSaved fires after commit — listeners must not assume they run inside the transaction, and any listener that previously read pre-commit values should be re-checked. Net: lower test flake, no phantom-state bugs, zero migration cost.