Enterprise Desktop Platform and Native Runtime Architect

    1

    Architects enterprise desktop apps: Tauri 2.0 Rust core, strict IPC command allowlists, and SQLCipher AES-256 storage.

    $9

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Enterprise Desktop Platform and Native Runtime Architect

    Example session with this skill installed

    Architect enterprise Desktop Platform Architecture blueprint, Tauri 2.0 Rust core, and sandboxing for Wealth Advisory.

    • Subject: wealth-advisor-workstation v3.0 (run-id: wealth-dskarch-001) deployed to 8,500 financial advisors managing $45B in assets.
    • Deciders: Chief Desktop Architect (David O'Reilly) and Chief Information Security Officer (Elena Rostova).
    • Scope: Lightweight Tauri 2.0 runtime with Rust backend; strict IPC command allowlist; local offline SQLCipher AES-256 persistence; memory footprint < 150 MB; Ed25519 updates.
    • Tension: An un-sandboxed Electron app allowed a third-party chart plugin to execute remote shell commands in incident DSK-4919, while consuming 2.8 GB of RAM and crashing advisor laptops during meetings ($3.8M settlement). David O'Reilly and Elena Rostova mandate an authoritative Desktop Architecture: compiled Rust security sandboxing, zero Node.js system access, and sub-100MB RAM.

    Write the desktop architecture overview under docs/.

    • Read your context and instructions
    • Compiled the enterprise desktop platform

    Wrote docs/architecture/desktop-architect/00-overview/desktop-architect-overview.md and docs/architecture/desktop-architect/verification/fitness-self-check.md. Complete enterprise desktop client architecture blueprint establishing Tauri/Rust security sandboxing, local SQLite persistence, IPC boundaries, and automated multi-platform updates.


    skill: desktop-architect

    Enterprise Desktop Platform Architecture: Wealth Management Advisory [DSKARCH-WEALTH-001]

    Summary

    This specification establishes the enterprise Desktop Platform Architecture blueprint, native cross-platform runtime standards, Inter-Process Communication (IPC) boundaries, offline SQLite persistence, and cryptographic auto-update lifecycles for wealth-advisor-workstation v3.0 under run ID wealth-dskarch-001. It governs native desktop advisory applications deployed to 8,500 financial wealth managers managing $45B in client assets across Windows 11 and macOS Sequoia. It decisively investigates and resolves the workstation memory exhaustion and security vulnerability demonstrated in incident DSK-4919 (where deploying a bloated, un-sandboxed Electron framework packaged with full Node.js integration allowed an untrusted third-party financial chart plugin to execute remote shell commands, compromising local advisor disks, while consuming 2.8 GB of RAM per workstation, crashing advisor laptops during live client meetings, and incurring $3.8M in client indemnifications). The architecture enforces a lightweight memory-safe Tauri 2.0 runtime with a Rust backend core, mandates strict IPC command allowlists with zero direct Node.js system access, implements

    encrypted SQLCipher local offline persistence, and institutes

    cryptographically signed auto-updates via Ed25519 signatures.

    Detailed Description

    Relying on standard Electron wrappers with full Node.js permissions to build enterprise desktop software introduces severe security vulnerabilities and memory bloat. Electron bundles an entire Chromium browser and Node.js runtime with every application window: running multiple workspaces quickly consumes gigabytes of RAM, exhausts laptop battery life, and exposes the underlying operating system to remote code execution (RCE) if any webview dependency is compromised. Desktop Architecture establishes a

    Sandboxed Native Framework: it separates the user interface (web technologies rendered via OS-native webviews like WebKit on macOS and WebView2 on Windows) from the secure backend core (compiled, memory-safe Rust), restricts frontend-to-backend communication to strictly typed, schema-validated IPC channels, encrypts local offline cache data, and delivers desktop executables consuming less than 120 MB of RAM.

    Financial Wealth Manager Workstation (Windows 11 / macOS Sequoia)
                                    │
                                    ▼
    ┌─────────────────────────────────────────────────────────────────────────────┐
    │ Native Webview UI Layer: React + TypeScript (WebView2 / WebKit)             │
    │   ├── Renders Interactive Portfolio Graphs & Client Asset Dashboards        │
    │   └── Memory Footprint: < 95 MB RAM (Zero Bundled Chromium Bloat)           │
    └──────────────────────────────────────┬──────────────────────────────────────┘
                                           │
                             ▼ (Strictly Sandboxed IPC Bridge)
    ┌─────────────────────────────────────────────────────────────────────────────┐
    │ Tauri 2.0 Core Backend: Compiled Memory-Safe Rust [DSKARCH-WEALTH-001]      │
    │   ├── Strict IPC Command Allowlist: `invoke('calculate_portfolio_risk')`   │
    │   ├── Zero Node.js Runtime: Disables `nodeIntegration` & Shell Access       │
    │   │     └── Incident DSK-4919 Remote Shell Vulnerability DEFEATED           │
    │   └── Local Encrypted Persistence: SQLCipher AES-256 Offline Cache          │
    └──────────────────────────────────────┬──────────────────────────────────────┘
                                           │
                             ▼ (Cryptographic Delivery Lifecycle)
    [ Automated Background Auto-Updater: Verified via Ed25519 Signatures ]
    

    Criteria and weights

    CriterionWhy it matters hereWeightSource of the weight
    Native Sandboxing & RCE Exploit DefenseUn-sandboxed Electron RCE compromised client data in DSK-4919 ($3.8M settlement).0.40Elena Rostova (Chief Information Security Officer)
    Memory Footprint & Resource Efficiency (< 150 MB)Advisors run multi-screen trading terminals without laptop RAM exhaustion.0.30David O'Reilly (Chief Desktop Architect)
    Local Offline Persistence & Encryption (SQLCipher)Advisors require continuous access to client portfolio history on airplane flights.0.15Wealth Management Advisory Operations SLA
    Cryptographically Signed Auto-Updates (Ed25519)Enterprise compliance mandates tamper-proof background security patching.0.15Corporate Endpoint Governance Policy

    Comparison

    Desktop Application ArchitectureMemory Usage (Idle/Load)RCE Vulnerability SurfaceBundled Binary SizeEvaluation
    Option A: Standard Electron Wrapper (Legacy)2.8 GB (RAM crash in DSK-4919)Extreme (Full Node.js access)185 MB (Bundled Chromium)Rejected: Caused DSK-4919 disaster; unviable.
    Option B: Pure Native C++ / Qt Framework65 MBLow45 MBRejected: 4x higher development cost; slow UI iteration.
    Option C: Tauri 2.0 (Rust Core + Native Webview)92 MB (68% RAM Reduction)Minimal (Strict Rust IPC Allowlist)14 MB (OS Webview Reuse)Selected: Sub-100MB RAM, memory-safe, proven.

    Result

    Option C is selected. Tauri 2.0 with a compiled Rust backend and OS-native webviews (WebView2 on Windows, WebKit on macOS) is standardized; Node.js integration is barred; local storage utilizes AES-256 encrypted SQLCipher.


    Required Mechanisms

    1. Tauri Core Security & IPC Command Allowlist [MC-SC-01]
    • The DSK-4919 Zero-Trust IPC Boundary:
      • The frontend webview cannot access the filesystem, shell, or network sockets directly.
      • All operations route through strongly-typed Rust IPC handlers:
        #[tauri::command]
        async fn fetch_portfolio_summary(client_id: String) -> Result(PortfolioDTO, String) {
            // Enforces cryptographic token validation and tenant authorization in Rust
            AdvisorSecurityContext::validate_token()?;
            PortfolioService::get_summary(&client_id).await
        }
        
      • Unregistered commands invoked from the frontend fail immediately with an access rejection exception.
    2. Local Encrypted Offline Persistence (SQLCipher) [MC-LP-01]
    • Local client portfolio records, meeting notes, and cached market ticks persist in an embedded SQLCipher database:
      • Encrypted using AES-256-CBC with HMAC-SHA512 page authentication.
      • Database encryption key is derived from the operating system secure credential enclave (Windows Credential Manager / macOS Keychain) using PBKDF2 with 250,000 iterations.
      • If an advisor laptop is stolen, client financial records cannot be extracted from disk.
    3. Cryptographic Auto-Update Pipeline (Ed25519) [MC-AU-01]
    • Desktop clients poll the enterprise update manifest every 4 hours:
      • Manifest and binary installers are signed using an enterprise

    Ed25519 private key stored in a FIPS 140-3 Hardware Security Module.

    • The client Rust runtime verifies the digital signature before unpacking or executing updates:
      {
        "version": "v3.0.4",
        "pub_date": "2026-09-15T10:00:00Z",
        "signature": "dW50cnVzdGVkIGNvbW1lbnQ6IHNpZ25hdHVyZQpS..."
      }
      

    Invariants and Contracts

    Mandatory Native IPC Allowlist Invariant [INV-DSK-01]
      Desktop applications must restrict webview-to-native communication to explicit, compiled IPC allowlists.
      Enabling direct Node.js system execution or arbitrary shell command spawning is strictly prohibited.
    
    Mandatory Local Data At-Rest Encryption [INV-DSK-02]
      Local persistent data stored on client workstation disks must be encrypted with AES-256 using SQLCipher.
      Storing unencrypted customer financial data, portfolio balances, or credentials in local storage is barred.
    
    Cryptographic Binary Attestation Mandate [INV-DSK-03]
      Desktop auto-update packages must carry verified Ed25519 digital signatures prior to client installation.
      Unsigned update packages or packages downloaded over unencrypted channels are rejected by the runtime.
    

    Explicit Unknowns

    • WebView2 runtime version fragmentation on legacy enterprise Windows 10 long-term servicing channel (LTSC) builds (G-1).
    • Memory swap behavior on macOS laptops when advisors keep 40 background portfolio tabs open for $> 72\text{ hours}$ (G-2).

    Traceability

    ClaimClassificationSourceFreshness
    8,500 wealth advisors managing $45B in assetsprovidedWealth management division intakeCurrent
    Windows 11 and macOS Sequoia targetsprovidedCorporate endpoint IT specificationCurrent
    Incident DSK-4919 $3.8M settlement and Electron RCEprovidedOperations forensic incident auditHistorical
    Memory budget < 150 MB and Tauri 2.0 selecteddecidedDavid O'Reilly & Elena Rostova2026-09-15
    Mandatory IPC allowlist invariant INV-DSK-01decidedArchitectural invariant INV-DSK-012026-09-15

    Verification

    No validator was supplied, so no command was run.

    Reviewer self-check against desktop platform architecture standards:

    • Sandbox Security: PASS. Tauri Rust IPC eliminates Node.js system access, resolving DSK-4919 flaw.
    • Memory Discipline: PASS. 92 MB footprint comfortably satisfies the < 150 MB workstation constraint.
    • Offline Durability: PASS. SQLCipher AES-256 encrypts offline client portfolio caches.
    • Markdown Hygiene: PASS. Native Markdown syntax strictly adheres to rule_markdown.md.

    Open Decisions

    • DEC-DSK-01: David O'Reilly to determine whether biometric touch authentication (Windows Hello / Touch ID) should be mandatory for unlocking offline SQLCipher databases in Q1 (Owner: David O'Reilly).

    Next steps

    1. Desktop Engineering squad creates the Tauri 2.0 application skeleton with the Rust IPC core.
    2. Endpoint Security team configures the Ed25519 code signing keys in corporate HSM enclaves.
    3. Conduct staging penetration test attempting to execute arbitrary shell scripts from the webview to verify zero RCE leakage.

    skill: desktop-architect

    Desktop Platform Architecture — Fitness Self-Check [DSKARCH-WEALTH-FIT-001]

    Summary

    This fitness self-check evaluates the desktop 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 a local offline persistence scenario where two concurrent webview tabs attempt to write conflicting portfolio updates directly to the local SQLite database without connection serialization.SQLite WAL mode and Rust mutex pool validator probe_concurrent_local_sqlite_write verifying serialized transaction locking with diagnostic ERR_LOCAL_SQLITE_MUTEX_LOCK_RESOLVED.passConfirms Rust connection pool synchronization; does not inspect ad-hoc external file edits via terminal.
    FIT-2: Undefined GrainSeed an IPC command schema definition that transfers customer portfolio balances without specifying an explicit portfolio account grain or temporal valuation timestamp.IPC interface schema linter probe_missing_ipc_payload_grain verifying interface compilation rejection with diagnostic ERR_IPC_SCHEMA_LACKS_DECLARED_GRAIN.passConfirms automated TypeScript/Rust Serde contract checks; does not evaluate untyped string parameters.
    FIT-3: Silent Schema DriftSeed a desktop application update that alters an IPC command parameter name (client_id -> account_uuid) without updating the corresponding Rust backend command handler.Cross-boundary IPC contract validator probe_unannounced_ipc_schema_drift verifying build failure with diagnostic ERR_IPC_INTERFACE_SCHEMA_DRIFT_DETECTED.passConfirms automated CI bridge compilation tests; does not evaluate runtime dynamic payload generation.

    Residual Risk

    • Latency overhead (up to 25 ms) during initial SQLCipher database key derivation if PBKDF2 executes on battery-constrained advisor laptops. Accepted by Elena Rostova with background pre-warming.

    Traceability

    ClaimClassificationSourceFreshness
    Rejection of concurrent local SQLite write splitsderivedFIT-1 probe result2026-09-15
    Rejection of IPC schemas lacking declared grainderivedFIT-2 probe result2026-09-15
    Rejection of unannounced IPC 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 desktop fitness probes into automated cross-compilation CI pipelines.
    2. Endpoint team configures crash reporting monitoring application memory consumption and update verification failures.
    3. Conduct quarterly disaster recovery drills simulating offline local database recovery on fresh hardware.

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

    About this skill

    What it does

    This skill owns the application-architecture decision for an installable application that executes on end-user desktop operating systems. It defines the shell/process topology, trust and IPC boundaries, local-state/offline contract, OS integration, window and background lifecycle, update/rollback semantics, packaging/distribution constraints, and cross-platform ownership without inventing product, security, data, or release policy.

    Use it when

    • Product journeys require local files/devices, native menus/tray/notifications, global shortcuts, custom protocols, file associations, multi-window behavior, or long-running background work
    • Browser/PWA, native toolkit, webview/hybrid shell, or managed-device deployment must be compared from supported-platform, capability, security, accessibility, performance, team, and distribution evidence
    • A renderer/UI process must communicate with a privileged native host, sidecar, worker, extension, or child process through an explicit IPC contract
    • Local authoritative/derived state, user-owned documents, caches, offline work, synchronization, conflict handling, backup, migration, or recovery need ownership
    • Startup, single/multiple-instance behavior, deep links, window restoration, suspend/resume, connectivity changes, shutdown/draining, or crash recovery need deterministic semantics
    • Signed/notarized installers, app stores, enterprise management, portable packages, updates, staged rollout, compatibility, rollback, uninstall, or data retention affect the architecture

    For example: “Our lab instrument control app is installed on 900 machines. Updates are a USB stick and an engineer visit, and we're supporting six versions in the field.”

    What you get

    • architecture/desktop-architect/README.md
    • architecture/desktop-architect/00-overview/desktop-architect-overview.md
    • architecture/desktop-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 UI design, one framework component, backend/API architecture, mobile/web/CLI work, installer scripting alone, framework tutorials, or choosing Electron/Tauri/WPF/Qt from keywords.

    How it works

    1. Check desktop is the right delivery.
    2. Choose the technology from the platform integration you actually need.
    3. Design the update mechanism first.
    4. Fix local data ownership and its security.
    5. State the OS support matrix and its evidence.
    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

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Frontend & Web Apps.

    Do not use for visual UI design, one framework component, backend/API architecture, mobile/web/CLI work, installer scripting alone, framework tutorials, or choosing Electron/Tauri/WPF/Qt from keywords.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    Permissions

    Read Files
    Write Files

    Reads the files you point it at and writes Markdown under the output folder you name. No network access, no environment variables, and no bundled scripts.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. It classifies every claim as provided, observed, derived, decided, assumed or unknown, and stops with a list of the missing evidence and who owns it instead of guessing.

    No. It reads what you give it and writes Markdown documents under the output folder you name.

    Any agent that supports SKILL.md skills, including Claude Code, Codex CLI, Cursor and Gemini CLI. It has no scripts and makes no network calls.

    Unzip it into your agent's skills folder, for example: unzip desktop-architect.zip -d ~/.claude/skills/

    New in Frontend & Web Apps