What problem does it solve? Web apps often feel slow because every interaction waits on network requests, producing spinners and lag even when latency is acceptable. This Skill provides architectural patterns to make web applications feel instant by reading from local state instead of the server. ## Core Features & Use Cases - Local-first data layer: Store the workspace in IndexedDB, hydrate into an in-memory observable store (e.g., MobX), and render UI from local state while syncing mutations in the background. - Optimistic updates: Apply mutations locally first with libraries like Tanstack Query or SWR, rolling back only if the server rejects. - Performance tactics: Per-package vendor chunking, modulepreload of critical chunks, service worker precaching, inlined critical CSS, variable font preloading, and GPU-only animations on transform and opacity. - Use Case: When users report your productivity app feels slow despite acceptable latency, apply these patterns to eliminate spinners, render cached data before authentication completes, and add keyboard-first interactions with a local command palette. ## Quick Start Review my web app's data flow and rendering pipeline, then apply local-first architecture patterns to eliminate loading spinners and make interactions feel instant.
