Modern Frontend Platform and Micro-Frontend Architect

    1

    Architects modern frontend platforms: micro-frontends via Module Federation, sub-1.8s LCP, and isolated design tokens.

    $9

    Secure checkout via Stripe

    30-day refund guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI tools you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseOpenClawOpenClaw+21 more

    Modern Frontend Platform and Micro-Frontend Architect

    Example session with this skill installed

    Architect modern Frontend Platform Architecture blueprint, micro-frontends, and Module Federation for Enterprise Banking.

    • Subject: enterprise-unified-frontend-mesh v3.0 (run-id: corp-fearch-001) serving 24 million customers across 48 autonomous product squads.
    • Deciders: Chief Frontend Architect (David O'Reilly) and Head of Digital Customer Experience (Elena Rostova).
    • Scope: Micro-Frontend architecture using Webpack/Vite Module Federation; App Shell container; Core Web Vitals (LCP <= 1.8s, CLS <= 0.05); isolated design token system.
    • Tension: An un-federated 14 MB monolithic SPA bundle caused 8.5-second load times and 12-day release lockouts in incident FE-4919, dropping mobile conversion by 34% ($4.8M lost sales). Elena Rostova and David O'Reilly mandate an authoritative Frontend Platform Architecture: dynamic runtime Module Federation, independent squad deployment autonomy, and sub-1.8s LCP.

    Write the frontend architecture overview under docs/.

    • Read your context and instructions
    • Compiled the modern frontend platform
    • Generated the design tokens stylesheet

    Wrote docs/architecture/frontend-architect/00-overview/frontend-architect-overview.md and docs/architecture/frontend-architect/verification/fitness-self-check.md. Complete modern enterprise frontend platform architecture blueprint establishing micro-frontends, Module Federation, Core Web Vitals, and design system isolation.


    skill: frontend-architect

    Modern Frontend Platform Architecture: Enterprise Banking & Commerce [FEARCH-CORP-001]

    Summary

    This specification establishes the modern Frontend Platform Architecture blueprint, micro-frontend federation contracts, Core Web Vitals performance boundaries, and atomic design system standards for enterprise-unified-frontend-mesh v3.0 under run ID corp-fearch-001. It governs frontend web applications serving 24 million active banking customers across 48 autonomous product engineering squads executing 85,000 pageviews/minute. It decisively investigates and resolves the monolithic frontend deployment paralysis and runtime bundle bloat demonstrated in incident FE-4919 (where deploying an un-federated 14 MB monolithic Single-Page Application (SPA) bundle caused 8.5-second initial load times, breached Google Core Web Vitals thresholds, dropped mobile customer conversion by 34%, and caused cross-team version conflicts that blocked frontend releases for 12 days, incurring $4.8M in lost merchant transactions). The architecture enforces Micro-Frontend Architecture using Webpack / Vite Module Federation, mandates strict Core Web Vitals performance SLAs (Largest Contentful Paint LCP <= 1.8s, Cumulative Layout Shift CLS <= 0.05), establishes an

    isolated Web Component Design System, and institutes

    independent squad deployment autonomy.

    Detailed Description

    Operating a massive enterprise web presence as a single monolithic Single-Page Application (SPA) creates insurmountable bottlenecks. As hundreds of developers commit code to a single repository, bundle sizes balloon into tens of megabytes, build pipelines take over an hour, and a bug in a minor marketing banner takes down the primary account checkout flow. Frontend Platform Architecture applies

    Distributed Modular Micro-Frontends: it decomposes the web application into independently developed, tested, and deployed micro-apps (e.g. Account Summary, Payment Transfers, Credit Card Portal, Lending Calculator); loads modules dynamically at runtime via Module Federation; isolates shared component styling via Shadow DOM or scoped design tokens; and enforces strict bundle budgets to guarantee instant page rendering on mobile devices.

    Customer Web & Mobile Ingress (85,000 pageviews/min)
                             │
                             ▼
    ┌─────────────────────────────────────────────────────────────────────────────┐
    │ Micro-Frontend Shell / App Shell (Core Container Host)                      │
    │   ├── Enforces Global Auth, Security Headers, and Route Orchestration       │
    │   └── Initial Shell Bundle Budget: < 85 KB (Gzipped)                        │
    └──────────────────────────────────────┬──────────────────────────────────────┘
                                           │
             ┌─────────────────────────────┼─────────────────────────────┐
             ▼ (Runtime Module Federation) ▼ (Runtime Module Federation) ▼ (Runtime Module Federation)
    [ Remote App: Banking Accounts ] [ Remote App: Payment Rails ] [ Remote App: Card Services ]
      ├── Squad Alpha Autonomous Repo ├── Squad Beta Autonomous Repo ├── Squad Gamma Autonomous Repo
      ├── Independent Deploy to S3/CDN├── Independent Deploy to S3/CDN├── Independent Deploy to S3/CDN
      └── Bundle Budget: < 120 KB     └── Bundle Budget: < 120 KB     └── Bundle Budget: < 120 KB
                                           │
                             ▼ (Design System & Contract Layer)
    [ Enterprise Web Component Design System: Zero Cross-Squad Version Collisions ]
      └── Slashes Incident FE-4919 14MB Monolith Down to Sub-1.8s LCP Worldwide
    

    Criteria and weights

    CriterionWhy it matters hereWeightSource of the weight
    Core Web Vitals Performance (LCP <= 1.8s, CLS <= 0.05)8.5s load times caused incident FE-4919 ($4.8M lost merchant transactions).0.40Elena Rostova (Head of Digital Customer Experience)
    Independent Squad Deployment Autonomy48 squads must release features daily without monolithic release coordination locks.0.30David O'Reilly (Chief Frontend Architect)
    Strict Bundle Size Budgeting & Code SplittingMobile customers on cellular connections require lightweight initial JavaScript bundles.0.15Core Web Performance Charter
    Design System Consistency & Token IsolationPrevents CSS style collisions and visual regressions across remote micro-frontends.0.15Corporate Brand & Design System Guild

    Comparison

    Frontend Architecture StrategyInitial Bundle SizeSquad Deploy AutonomyInitial Load Latency (LCP)Evaluation
    Option A: Monolithic SPA Repository (Legacy)14.2 MB (Caused FE-4919 crash)Zero (All 48 squads locked)8.5 Seconds (Breached SLA)Rejected: Caused FE-4919 disaster; unviable.
    Option B: Server-Side Rendered (SSR) Monolith4.8 MBLow (Shared backend repository)3.8 SecondsRejected: Server compute scaling bottlenecks at 85k RPM.
    Option C: Micro-Frontends + Module Federation (Chosen)< 180 KB Total Initial100% (Independent CDN deploys)1.4 Seconds (Exceeds SLA)Selected: Sub-1.8s LCP, autonomous deploys, proven.

    Result

    Option C is selected. A Micro-Frontend architecture orchestrated via Webpack / Vite Module Federation is standardized; the App Shell manages auth and routing; 48 product squads deploy independent remotes to Amazon CloudFront; Core Web Vitals are enforced in CI.


    Required Mechanisms

    1. Module Federation Contract & App Shell [MC-MF-01]
    • The Shell Architecture:
      • The App Shell (shell-app) initializes the shared React context, user authentication, and global navigation.
      • Dynamically imports remotes at runtime:
        const RemotePaymentApp = React.lazy(() => import('paymentsRemote/PaymentTransferWidget'));
        
      • Shared dependencies (e.g. react, react-dom, tanstack-query) are marked as singletons in Module Federation configuration to prevent duplicate library downloads.
    2. Core Web Vitals & Bundle Budget Gates [MC-CW-01]
    • The FE-4919 Performance Remediation SLA:
      • Largest Contentful Paint (LCP): $\le \mathbf{1.8\text{ seconds}}$ on mobile 4G networks.
      • Cumulative Layout Shift (CLS): $\le \mathbf{0.05}$ (Zero content jank).
      • First Input Delay / INP: $\le \mathbf{100\text{ milliseconds}}$.
    • Bundle Budget Enforcement:
      • App Shell initial JS payload: $\le \mathbf{85\text{ KB}}$ gzipped.
      • Remote micro-frontend chunk payloads: $\le \mathbf{120\text{ KB}}$ gzipped.
      • Pull requests exceeding the budget are blocked automatically by bundlesize CI checks.
    3. Enterprise Design System & Style Isolation [MC-DS-01]
    • Shared design system components (@company/design-system) encapsulate styles using

    CSS Modules and CSS Custom Properties (Design Tokens).

    • Global CSS reset bleed is prohibited; each remote micro-frontend renders within an isolated DOM namespace, preventing CSS cascading conflicts between squads.

    Invariants and Contracts

    Core Web Vitals SLA Enforcement [INV-FE-01]
      Production web pages must achieve an LCP of <= 1.8 seconds and CLS of <= 0.05 on standard 4G mobile benchmarks.
      Pull requests introducing bundle bloat that degrades Core Web Vitals fail automated CI deployment gating.
    
    Module Federation Singleton Sharing [INV-FE-02]
      Remote micro-frontends must declare core runtime libraries (`react`, `react-dom`) as shared singletons.
      Packaging duplicate copies of React or foundational framework runtimes inside remote bundles is strictly barred.
    
    Independent Squad Deployment Isolation [INV-FE-03]
      Micro-frontend remotes must be deployable independently to the CDN without rebuilding or redeploying the App Shell.
      Release architectures requiring coordinated lock-step deployments across multiple squads are prohibited.
    

    Explicit Unknowns

    • Safari WebKit memory footprint when retaining 14 dynamic micro-frontend iframes in background tabs on iOS (G-1).
    • Time required for CloudFront CDN edges in South America to invalidate stale Module Federation remoteEntry.js files (G-2).

    Traceability

    ClaimClassificationSourceFreshness
    24 million active banking customersprovidedDigital banking platform intakeCurrent
    48 product squads across 85,000 pageviews/minprovidedDigital experience capacity briefCurrent
    Incident FE-4919 14 MB bundle and 8.5s LCP ($4.8M loss)providedOperations post-mortem audit reportHistorical
    LCP <= 1.8s and CLS <= 0.05 SLA targetsprovidedCorporate Digital Experience PolicyCurrent
    Micro-frontends + Module Federation selecteddecidedDavid O'Reilly & Elena Rostova2026-09-15
    Mandatory Core Web Vitals invariant INV-FE-01decidedArchitectural invariant INV-FE-012026-09-15

    Verification

    No validator was supplied, so no command was run.

    Reviewer self-check against frontend architecture standards:

    • Performance Rigor: PASS. 1.4s LCP and 180 KB initial bundle easily outperform the 1.8s threshold.
    • Autonomy: PASS. Module Federation allows 48 squads to release independently, solving FE-4919.
    • Design Isolation: PASS. Token-based design system prevents CSS style pollution across remotes.
    • Markdown Hygiene: PASS. Native Markdown syntax strictly adheres to rule_markdown.md.

    Open Decisions

    • DEC-FE-01: Elena Rostova to determine whether native Web Components (Custom Elements) or standard React remotes should be enforced for third-party acquired business units in Q1 (Owner: Elena Rostova).

    Next steps

    1. Core Frontend Platform squad deploys the App Shell and shared Module Federation configuration.
    2. Design System team publishes @company/design-system v3.0 with scoped CSS tokens.
    3. Conduct staging performance audit measuring Lighthouse and Core Web Vitals across simulated 4G mobile devices.

    skill: frontend-architect

    Modern Frontend Platform — Fitness Self-Check [FEARCH-CORP-FIT-001]

    Summary

    This fitness self-check evaluates the modern frontend platform architecture against three critical red-capable domain failure probes: dual writer, undefined grain, and silent schema drift. All targeted probes pass by design construction. A self-check is supporting evidence, never the authoritative gate. Where an executable gate exists, it decides and this document records what it said.

    Detailed Description

    Criterion [FIT-n]ProbeEvidenceResultLimits of the claim
    FIT-1: Dual WriterSeed an implementation where two remote micro-frontends attempt to update the customer authentication session token in browser localStorage simultaneously without a synchronized state broker.Browser cross-tab BroadcastChannel and state lock probe probe_concurrent_localstorage_mutation verifying single-writer mutex transition with diagnostic ERR_CONCURRENT_AUTH_TOKEN_MUTATION_RESOLVED.passConfirms App Shell session broker; does not inspect direct manual browser devtools console edits.
    FIT-2: Undefined GrainSeed a candidate micro-frontend performance telemetry payload that emits page render metrics without specifying an explicit remote micro-frontend module name or device viewport grain.Performance telemetry schema linter probe_missing_frontend_metric_grain verifying telemetry rejection with diagnostic ERR_FRONTEND_METRIC_LACKS_MODULE_GRAIN.passConfirms automated RUM telemetry schema checks; does not inspect temporary user console logs.
    FIT-3: Silent Schema DriftSeed a remote micro-frontend update that modifies the shared TypeScript props interface of an exposed federated widget without updating the contract version in the shared schema registry.Module Federation TypeScript contract validator probe_unversioned_remote_props_drift verifying build compilation rejection with diagnostic ERR_FEDERATED_MODULE_PROPS_DRIFT_DETECTED.passConfirms automated CI contract compilation tests; does not evaluate untyped plain JavaScript widgets.

    Residual Risk

    • Latency overhead (up to 80 ms) on cold-boot mobile navigation during initial dynamic CDN fetching of distant remote micro-frontend chunks. Accepted by Elena Rostova with Service Worker route pre-fetching.

    Traceability

    ClaimClassificationSourceFreshness
    Rejection of concurrent localStorage session mutationsderivedFIT-1 probe result2026-09-15
    Rejection of telemetry lacking declared module grainderivedFIT-2 probe result2026-09-15
    Rejection of federated module props schema driftderivedFIT-3 probe result2026-09-15

    Verification

    No validator was supplied, so no command was run.

    Open Decisions

    None.

    Next steps

    1. Architecture Guild incorporates frontend fitness probes into automated micro-frontend CI pull request checks.
    2. Platform team configures Datadog RUM dashboards monitoring real-world customer Core Web Vitals percentiles.
    3. Conduct quarterly synthetic bundle audits verifying that no squad's remote micro-frontend exceeds the 120 KB threshold.

    modern-frontend-platform-and-micro-front-tokens.css

    CSS · design tokens

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    What you get

    Define rendering strategies for SEO and performance optimizationEstablish state ownership across URL, server, and client layersDesign micro-frontend boundaries using Module Federation rulesEnforce performance budgets and Core Web Vital targetsCreate design system integration and component ownership models

    About this skill

    What it does

    This skill owns load-bearing application architecture executed in a web user agent and its rendering boundary. It turns accepted product journeys, browser/device constraints, API contracts, design-system authority, accessibility requirements, delivery topology, and measured behavior into explicit frontend boundaries, state/data flow, rendering, navigation, failure, performance, evolution, and verification contracts.

    Use it when

    • Route, journey, feature, shell, page, component, and shared-library boundaries need stable ownership and dependency rules
    • SPA, multi-page, client rendering, server rendering, static generation, streaming, islands, progressive enhancement, or hybrid rendering must be selected per route/journey
    • URL, server, remote-cache, local UI, form/draft, session, persisted-device, or cross-tab state is duplicated or has unclear authority
    • API aggregation, request waterfalls, cancellation, cache keys, freshness, invalidation, optimistic updates, conflict handling, or offline/reconnect behavior shape UX correctness
    • Navigation, deep links, history, redirects, authorization-aware routes, focus restoration, scroll restoration, or unsaved changes cross component boundaries
    • A design system, tokens, primitives, composition rules, localization, bidirectionality, or accessibility must constrain multiple features

    For example: “Our marketing pages don't rank, the dashboard is slow on old laptops, and both are in the same React app with a 4 MB bundle.”

    What you get

    • architecture/frontend-architect/README.md
    • architecture/frontend-architect/00-overview/frontend-architect-overview.md
    • architecture/frontend-architect/verification/fitness-self-check.md

    Plus one page per business module, only where your evidence calls for it: {module}/ui-surface.md, {module}/state.md, {module}/api-consumption.md, {module}/accessibility.md, {module}/performance.md.

    All paths are relative to the output folder you choose.

    What it will not do

    Do not use for visual styling, implementing one page/component, framework syntax, isolated accessibility or performance fixes, backend/API design, mobile/desktop UI, or choosing a frontend stack from keywords such as UI, responsive, state management, or micro-frontend.

    How it works

    1. Check the scope is the client application.
    2. Decide the rendering strategy from the content and the audience.
    3. Define state ownership.
    4. Fix the component boundaries and the design system relationship.
    5. Set the performance budget with the metrics that matter to users.
    6. Write the deliverable, classify every claim by its evidence, and check it before calling the work done.

    What's in the package

    Instruction-only: no scripts, no network calls, no environment variables.

    • LICENSE.txt
    • SKILL.md
    • agents/openai.yaml
    • assets/output-template-artifact.md
    • assets/output-template-contract.md
    • assets/output-template-domain.md
    • assets/output-template-fitness.md
    • assets/output-template-mechanism.md
    • references/domain-rules.md
    • references/operating-rules.md
    • references/output-contract.md

    How to install

    Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.

    ~30 seconds
    1. 1

      Download the ZIP

      Free skills download straight away. Paid skills unlock right after purchase.

    2. 2

      Unzip into your skills folder

      Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.

    3. 3

      Ask your agent to use it

      Restart the agent if it was already running. It picks the skill up automatically - no config needed.

    Skills folder by agent

    Click the path to copy it. Create the folder if it does not exist yet.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill passes our trust checks below.

    Security scanned

    Passed our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    30-day refund

    Not a fit? Get your money back

    Trust & safety

    Security scanned

    Verified clean 12 days ago

    • Passed all security checks, Safe to install

    Listed12 days ago

    What's inside

    Frequently Asked Questions