production-performance-and-efficiency

Files (1)

• SKILL.md

SKILL.md (first lines)

# Production Performance & Efficiency Standards

## Core Principles
Build for correctness, security, reliability, maintainability, and efficiency. Minimize unnecessary computation, data transfer, database operations, network requests, and user waiting time.

## Rules
1. **Minimize Resource Overhead**: Avoid redundant work, duplicate queries, payload bloat, unnecessary re-renders, or blocking main-thread computations.
2. **Justified Optimization Patterns**: Apply pattern techniques only when justified by architecture or scale requirements:
   - **Caching**: Store expensive compute or frequent immutable queries.
   - **Batching & Concurrency**: Group operations and run independent tasks concurrently.
   - **Lazy Loading & Pagination**: Defer non-critical resource loading and paginate large datasets.
   - **Background Processing**: Offload long-running operations off the critical user-waiting path.
3. **No Premature Optimization**: Do NOT introduce complex optimizations without a demonstrated bottleneck, measurable impact, or clear architectural reason.
4. **Validation & Metrics**: Verify performance improvements using application-specific tests, metrics, or performance audits (e.g., Lighthouse / DevTools).

---

## 3. Just-In-Time (JIT) Telemetry & Sub-Module Isolation Protocol
When designing complex dashboards, administrative portals, or multi-tab screens:
1. **Single-Page In-Memory Living Shell (Zero Hard Multi-Page Roundtrips)**:
   - Strictly forbid splitting operator, merchant, or admin dashboards into separate hard Next.js routes (`/admin/tab-a`, `/store/tab-b`) with `force-dynamic` that destroy DOM state and force multi-hop SSR roundtrips on every tab tap.
   - Every operator dashboard MUST be architected as a single-page in-memory living shell where tab switches occur in **0ms** via URL hash (`#module`), preserving operator search terms, filters, and scroll positions in client memory (`loadedModules`).
2. **Sub-30ms Instant Mount Telemetry**:
   - Initial server-rendered payload or client mount must strictly fetch scalar aggregate counts (`count: 'exact'`, `head: true`, or specialized summary RPCs).
   - Strictly forbid dumping bulk entity rows, large catalog tables, or deep relational joins on root screen load.
3. **On-Demand JIT Sub-Module Fetching & Memory Caching**:
   - Defer sub-module, tab, drawer, and modal data loading until direct user activation.
   - Maintain client module state (`loadedModules`) to cache fetched results in memory—guaranteeing 0ms subsequent tab navigation without redundant database roundtrips.
   - Always expose an explicit "Sync Live / Refresh" trigger allowing operators to invalidate and reload the active module on demand.
4. **URL State & Navigation Reversibility**:
   - Synchronize active sub-modules with the URL (hash `#module` or search param `?tab=module`).
   - Ensure bookmarking, browser back/forward buttons, and page refreshes preserve the exact operator context.
5. **Domain-Isolated Sub-Component Architecture**:
   - Never build monolithic multi-tab components (>300 lines combining distinct operational domains).
   - Decompose every tab, filter strip, modal, and metric bar into single-responsibility sub-components under a dedicated `/components/` sub-directory.

From hemanth17r/ahr-workflow, MIT.

More from hemanth17r/ahr-workflow