# Implementation Plan: Central Tenant Operability Policy

**Branch**: `148-central-tenant-operability-policy` | **Date**: 2026-03-17 | **Spec**: [specs/148-central-tenant-operability-policy/spec.md](./spec.md)
**Input**: Feature specification from `/specs/148-central-tenant-operability-policy/spec.md`

**Note**: This template is filled in by the `/speckit.plan` command. See `.specify/scripts/` for helper scripts.

## Summary

Consolidate existing tenant lifecycle and selector semantics into one actor-aware, lane-aware tenant operability authority that returns reusable structured decisions instead of scattered booleans. The implementation will evolve the existing `TenantOperabilityService` and related tenant-support types into a central policy boundary consumed by selector membership, remembered-context revalidation, tenant action catalogs, tenant-bound route checks, onboarding workflow surfaces, and canonical workspace record viewers.

This is a support-layer hardening feature, not a persistence or UI-redesign feature. The plan therefore focuses on extending the current support seams already present in `app/Services/Tenants`, `app/Support/Tenants`, `app/Support/OperateHub`, and `app/Support/Workspaces`, then migrating the highest-risk consumers first: choose-tenant and context persistence, `OperateHubShell`, `TenantActionPolicySurface`, `TenantResource`, onboarding workflow routes, and `TenantlessOperationRunViewer`.

## Technical Context

**Language/Version**: PHP 8.4.15  
**Primary Dependencies**: Laravel 12, Filament 5, Livewire 4, Pest 4, existing support-layer helpers such as `UiEnforcement`, `CapabilityResolver`, `WorkspaceContext`, `OperateHubShell`, `TenantOperabilityService`, and `TenantActionPolicySurface`  
**Storage**: PostgreSQL plus existing session-backed workspace and remembered-tenant context; no schema change planned for the first implementation slice  
**Testing**: Pest 4 feature, unit, and Filament or Livewire-focused tests run through Laravel Sail  
**Target Platform**: Laravel Sail web application serving the Filament admin panel and workspace-canonical `/admin` routes
**Project Type**: Laravel monolith web application  
**Performance Goals**: Keep operability evaluation synchronous, deterministic, and DB or session-backed only; add no render-time external calls; avoid material query-count regressions on the chooser, tenant detail, and canonical operations viewer  
**Constraints**: Preserve Spec 143 lifecycle semantics, Spec 144 canonical-viewer authority, Spec 145 action taxonomy, Spec 146 badge centralization, and Spec 147 selector semantics; keep Livewire v4 and Filament v5 compliance; leave panel-provider registration unchanged in `bootstrap/providers.php`; keep `TenantResource` globally searchable only through its existing View or Edit pages; no new asset pipeline or `filament:assets` change; destructive actions remain `->action(...)->requiresConfirmation()`; 404 versus 403 semantics must stay explicit and server-side  
**Scale/Scope**: One central support-layer policy boundary plus adoption across the workspace shell, choose-tenant flow, remembered-context handling, tenant management resource, onboarding workflow surfaces, canonical operation viewer, and focused regression suites under `tests/Feature` and `tests/Unit`

## Constitution Check

*GATE: Must pass before Phase 0 research. Re-check after Phase 1 design.*

**Pre-Phase 0 Gate: PASS**

- Inventory-first: PASS. This feature does not change inventory, snapshot, or backup ownership and only hardens policy evaluation over existing records.
- Read/write separation: PASS. No new remote writes, queued workflows, or lifecycle mutation mechanics are introduced. Existing archive and restore actions remain confirmed and audited where already implemented.
- Graph contract path: PASS. No Microsoft Graph calls are added or changed.
- Deterministic capabilities: PASS. Capability checks remain routed through the existing canonical capability registry and resolver services.
- RBAC-UX planes: PASS. This feature remains in the admin `/admin` plane. Cross-plane behavior is unchanged. Non-members stay 404, in-scope capability denial stays 403.
- Workspace isolation: PASS. Workspace remains the primary operating boundary and remembered tenant context remains workspace-scoped convenience state only.
- Tenant isolation: PASS. Tenant-linked routes and canonical viewers must still enforce workspace membership plus tenant entitlement where applicable.
- Destructive confirmation: PASS. No new destructive actions are added. Existing `Archive` and `Restore` remain `->action(...)->requiresConfirmation()` and capability-gated.
- Global search safety: PASS WITH REQUIRED ADOPTION. `TenantResource` is already globally searchable and has View and Edit pages, satisfying Filament v5 global search requirements. The implementation must ensure selector eligibility is not incorrectly reused as the only discoverability rule.
- Run observability and Ops-UX: PASS. No new `OperationRun` creation, lifecycle transition, notification, or summary-count behavior is introduced.
- Badge semantics (BADGE-001): PASS. Lifecycle and status presentation stay centralized through existing badge and lifecycle presentation helpers.
- UI naming (UI-NAMING-001): PASS. Operability explanations will reuse existing domain terms such as `Archive`, `Restore`, `Resume onboarding`, `selected tenant`, `route tenant`, and `run tenant`.
- Filament Action Surface Contract: PASS. In-scope Filament surfaces already exist; this feature changes their decision authority, not their required action inventory. Any touched lifecycle mutation remains confirmation-required and audit-aware.
- Filament UX-001: PASS. No layout redesign is planned. Existing pages keep their current layout contracts while consuming central policy outcomes.
- Asset strategy: PASS. No new Filament or front-end assets are planned, so no deployment change to `php artisan filament:assets` is required.

**Post-Phase 1 Re-check: PASS**

- The design extends existing support-layer abstractions instead of adding a parallel policy subsystem.
- No database migration, queue job, Graph contract change, panel change, or asset build change is required for the first implementation slice.
- Livewire v4 and Filament v5 compliance remain intact, and provider registration continues to live in `bootstrap/providers.php`.
- `TenantResource` remains globally searchable through existing View or Edit pages; the design only changes search scoping inputs, not the resource’s page contract.

## Project Structure

### Documentation (this feature)

```text
specs/148-central-tenant-operability-policy/
├── plan.md
├── research.md
├── data-model.md
├── quickstart.md
├── contracts/
│   └── tenant-operability-policy.openapi.yaml
└── tasks.md
```

### Source Code (repository root)

```text
app/
├── Filament/
│   ├── Pages/
│   │   ├── Operations/
│   │   ├── Workspaces/
│   │   └── ChooseTenant.php
│   └── Resources/
│       └── TenantResource.php
├── Http/
│   └── Controllers/
├── Models/
│   ├── OperationRun.php
│   ├── Tenant.php
│   ├── TenantOnboardingSession.php
│   ├── User.php
│   └── UserTenantPreference.php
├── Policies/
├── Services/
│   ├── Auth/
│   ├── Onboarding/
│   └── Tenants/
└── Support/
    ├── Filament/
    ├── Middleware/
    ├── OperateHub/
    ├── Rbac/
    ├── Tenants/
    └── Workspaces/
routes/
└── web.php
resources/
└── views/
    └── filament/
tests/
├── Feature/
│   ├── Operations/
│   ├── Onboarding/
│   ├── Rbac/
│   └── Filament/
└── Unit/
    └── Tenants/
```

**Structure Decision**: Use the existing Laravel monolith and strengthen the support layer rather than creating a new policy module outside the current architecture. The implementation seam stays inside `app/Services/Tenants` and `app/Support/Tenants`, with adoption routed through `WorkspaceContext`, `OperateHubShell`, existing Filament resources and pages, and existing auth or UI enforcement helpers.

## Phase 0 Research Summary

- The repo already contains a partial centralization layer, but it is not sufficient for Spec 148. `TenantOperabilityService` currently returns lifecycle-derived booleans and only accepts a `Tenant`; it does not accept actor, lane, linked-record context, or capability context.
- `TenantActionPolicySurface` is already centralizing labels and action descriptors, which means the right direction is to feed it richer operability outcomes rather than replace it with another action-assembly system.
- `OperateHubShell` currently mixes route-authoritative tenant resolution, Filament tenant state, and remembered tenant fallback. Its entitlement logic uses `TenantPageCategory`, but outside tenant-bound routes it still relies on `canSelectAsContext`, which is too narrow for administrative discoverability and canonical linked-record handling.
- `ChooseTenant` and `SelectTenantController` duplicate selector-eligibility checks and persistence flow. These should become thin consumers of the shared operability policy and shared remembered-context validation path.
- `TenantResource::getGlobalSearchEloquentQuery()` currently reuses `applySelectableScope()`, which means selector eligibility is already influencing discoverability. Spec 148 needs a clear separation between selector lane eligibility and administrative discoverability.
- Route safety foundations are already present. `Tenant::resolveRouteBinding()` loads soft-deleted tenants for external IDs, `TenantPageCategory` already classifies canonical and onboarding routes, and `TenantlessOperationRunViewer` already treats tenant mismatch as informational. The feature should formalize and centralize the same semantics instead of re-deriving them locally.
- Laravel documentation confirms the intended 404 versus 403 split should use authorization responses such as `Response::denyAsNotFound()` for non-members or non-entitled access and normal Gate authorization for in-scope capability denial.
- Filament documentation confirms destructive actions must remain `->action(...)->requiresConfirmation()` and that URL-only actions cannot be confirmation-modal actions. Existing lifecycle actions already align with this and the implementation should preserve that contract.

## Phase 1 Design

### Implementation Approach

1. Evolve the existing operability seam into the single authority.
   - Keep `TenantOperabilityService` as the implementation anchor, but expand it from tenant-only booleans into a central policy surface that accepts explicit context.
   - Preserve backward compatibility temporarily by providing adapter methods for existing consumers while migrating them to structured outcomes.

2. Add explicit lane-aware and reason-aware domain objects.
   - Introduce or extend support-layer types for interaction lane, structured input context, structured outcome, and reason codes.
   - Keep these types in `app/Support/Tenants` so they remain close to existing lifecycle and page-category enums.
   - `TenantOperabilityContext` carries normalized record, lane, route, and workflow inputs supplied by the consumer, while `TenantOperabilityService` resolves live membership, entitlement, capability, and archived-state facts through existing authoritative helpers so callers do not precompute authorization truth.

3. Separate operability from authorization while composing them intentionally.
   - Operability decides whether the requested behavior is meaningful for lifecycle and lane.
   - Authorization remains server-side truth for membership, entitlement, and capability.
   - The structured outcome must distinguish at least wrong lane, lifecycle mismatch, non-selector eligibility, missing capability, and non-entitled access.

4. Centralize consumer adoption around the highest-risk surfaces first.
   - Selector construction and remembered-context validation.
   - `OperateHubShell` active-tenant resolution.
   - `TenantActionPolicySurface` and `TenantResource` action visibility.
   - Onboarding workflow resumability checks.
   - Canonical viewer tenant-linked affordance and tenant-bound route support checks.

5. Split selector eligibility from discoverability.
   - Replace the current assumption that active-selector membership and administrative visibility are the same semantic decision.
   - Create distinct policy questions for standard selector eligibility, administrative discoverability, tenant-bound viewability, and canonical reference validity.

6. Preserve route-authority semantics from Specs 144 and 147.
   - Tenant-bound pages remain valid from route tenant plus entitlement.
   - Canonical workspace viewers remain valid from the workspace-owned record plus entitlement.
   - Selected tenant context remains a filter or convenience input only.

7. Keep the first slice schema-free and asset-free.
   - No new persistent table is required for structured outcomes or reason codes.
   - No new panel, provider registration, or Filament asset work is planned.

### Planned Workstreams

- **Workstream A: Operability core model**
  Extend `TenantOperabilityService` and `TenantOperabilityDecision` into a central lane-aware policy boundary with structured inputs and reasoned outcomes.

- **Workstream B: Selector and remembered-context adoption**
  Refactor `ChooseTenant`, `SelectTenantController`, and `WorkspaceContext` to use the new operability contract for standard selector membership and remembered-context validity.

- **Workstream C: Shell and route adoption**
  Update `OperateHubShell`, route-adjacent middleware, and tenant-page classification consumers so administrative, tenant-bound, and canonical-record lanes request the right operability question instead of reusing selector booleans.

- **Workstream D: Action-surface adoption**
  Refactor `TenantActionPolicySurface` and `TenantResource` lifecycle actions to consume action-specific operability outcomes and reason codes while preserving `UiEnforcement`, confirmation, and audit behavior.

- **Workstream E: Discoverability and global-search hardening**
  Separate administrative discoverability from standard selector eligibility for `TenantResource`, choose-tenant, and any shared tenant-picking or search surfaces in scope.

- **Workstream F: Regression hardening**
  Add focused Pest coverage for lifecycle state, lane differences, selector membership, remembered-context invalidation, canonical viewer safety, tenant-bound route safety, lifecycle-action eligibility, and capability-denied versus lifecycle-denied distinctions.

### Testing Strategy

- Add unit tests for the central operability service covering each lifecycle across each lane and asserting structured reason codes for denied outcomes.
- Add unit tests for selector eligibility, remembered-context validity, tenant-bound viewability, canonical linked-record viewability, archive eligibility, restore eligibility, and resume-onboarding eligibility.
- Add focused feature tests for `ChooseTenant` and `SelectTenantController` to prove only active-lane eligible tenants can be selected.
- Add focused feature tests for `WorkspaceContext` or route-level consumers to prove stale remembered context is cleared when the tenant becomes ineligible, missing, or out of workspace.
- Add focused feature tests for `/admin/tenants/{tenant}` showing onboarding and archived tenants remain route-valid when authorized even though they are not selector-eligible.
- Add focused feature tests for `/admin/operations/{run}` showing canonical viewer legitimacy does not depend on selected tenant context.
- Add or update tests around `TenantActionPolicySurface` and `TenantResource` to prove `Archive`, `Restore`, and `Resume onboarding` remain lifecycle-safe and capability-aware.
- Add focused tests proving onboarding-completion plus readiness or verification affordances resolve from the central operability layer with capability-aware and lifecycle-aware denial reasons.
- Add focused search or resource-query tests proving tenant global search and administrative discoverability do not collapse back to selector-only semantics.
- Run the minimum focused Pest suite through Sail; no full suite is required for planning artifacts.

## Complexity Tracking

No constitution violations or exceptional complexity are planned at this stage.