REQUEST - Senior Full-Stack Developer (NON GREENFIELD)

Revision Note — v1.4 (identity & tracking — a first-class citizen). Made explicit that a service_request is a first-class object in the shared operational object model (Library/Board v1.5 §2), carrying the platform-standard identity ( uuid VARCHAR(36) + human-readable reference_id , exactly as organizations/hotels do per migration 010), tenant scope, lifecycle, immutable provenance, and relationship links — defined once, reused, never a parallel identity or tracking scheme. Added §11A Tracked with everything else — toward outcomes: requests correlate to the guest stay, room/asset, property, and time so reactive-work performance reports alongside all operational data and rolls up toward guest-experience outcomes (OHI). Object model, audit, and milestones updated. Revision Note — v1.3 (notifications & delivery, reconciled to Communications Layer v4). Added §7B Notifications & delivery — an event-based model where Requests emits its lifecycle events and subscribes to Comms intent events, never reaching into the Comms layer (Comms v4 §23 anti-rewrite spine). Channels: live voice announce, push via the Thin Native Container (Comms §20), in-app, email, SMS fallback — with delivery owned by Comms/the container, not Requests. Reuses Comms addressing resolvers (presence / department / position / area / board) instead of inventing routing. Adds a configurable settings matrix (event × channel × role, org/property defaults, shift-aware quiet hours, escalation ladder, safety override). §9 corrected: Comms transcribes + classifies + emits the intent event; Requests is the consumer (the one Comms v4 says is "built later in Agenda/PDEE"), with acknowledge ↔ request correlation so a spoken "taking it" can't drift from an open request. Dependencies, milestones, audit updated. Revision Note — v1.2 (communication intelligence + share + view receipts). Per direction, the request's communication is treated as a first-class data source — not just a thread to read later. Added: share (give someone visibility without making them an owner — a viewer grant via the existing chat participant role); view receipts (delivered → viewed → acknowledged, reusing messages_seen + last_read_at , with unviewed-by-assignee surfaced for accountability); voice notes transcribed into the searchable thread; and Intelligence over the thread (§8) — theme/asset extraction ("TV remote again"), sentiment/guest-impact flags, thread summaries for handoff, and resolution-knowledge capture that mines how a request was fixed and drafts it into the Library. All advisory, tenant-scoped, and functional with Intelligence disabled. Revision Note — v1.1 (collaboration added). A gap was caught: v1.0 covered evidence-on-close but not the live, day-to-day team interaction on a request — comments, responses, adding people, and attaching photos/files mid-flight — which the legacy system had. §7A is added to specify request collaboration, built entirely on existing infrastructure: the tie-message thread pattern ( message_model ), the chat participant model ( Chat_model — add/remove people, search), and the governed media spine ( Media_attachment / Media_file / Media_link ). No second chat system is built. Object model, dependencies, milestones, and audit updated to match. Revision Note — v1.0 (code-audit reconciliation). The develop branch was audited against this scope. Most dependencies check out and are richer than assumed; several findings changed this document so the developer builds onto existing infrastructure instead of reinventing it. Confirmed present (reuse, don't rebuild): the Foundations spine — organization , hotels , hotel_branches (hotel locations), and user_hotel_location_assignments + user_active_hotel_location for portfolio scoping; the inventory_usage / inventory_adjustment / inventory_restock ledger (a mature subsystem); the NotificationService library (the single notification entry point — the legacy ticket even disabled its own writes in favor of it); the ai_governance_* layer plus ai_structured_output_model (ideal for voice → structured draft); per-user language preferences + the translation layer; immutable audit_logs triggers. Findings that changed the scope: (1) scoping resolves through hotel_location_id (org → hotel → location), not hotel_id alone — corrected