Article Type: Concept
Audience: Enterprise Administrators, Application Administrators, Application Designers, Developers, Partners
Module: Fuuz Platform
Applies to Versions: 2026.2+
Fuuz publishes a curated set of Claude AI skills that provide deep, domain-specific knowledge of the Fuuz Industrial Operations Platform (fuuz.com). These skills are installable packages for the Anthropic Claude AI assistant that enable Claude to generate production-ready Fuuz artifacts — data models, data flows, screen JSON, importable packages, and industrial operations patterns — without requiring specialized knowledge of the underlying technologies such as GraphQL, MongoDB, React, RabbitMQ, Stimulsoft, FusionCharts, or FontAwesome.
The Fuuz Claude Skills are openly available to customers, partners, and the broader Fuuz community. Fuuz regularly updates these skills based on platform evolution and community contributions, creating a shared knowledge base that benefits every organization leveraging the Fuuz platform.
The Fuuz Claude Skills are organized as a layered suite. A shared platform reference skill provides cross-cutting knowledge, while specialized builder and domain pattern skills handle specific areas of the Fuuz platform. Together, the suite covers the complete application development lifecycle on Fuuz: schema design, server-side logic, user interface creation, package generation, and industrial domain patterns.
┌──────────────────────┐
│ fuuz-platform │
│ (shared reference) │
└──────────┬───────────┘
│
┌────────────────────┼────────────────────┐
▼ ▼ ▼
┌──────────────────┐ ┌──────────────────┐ ┌──────────────────┐
│ fuuz-schema │ │ fuuz-flows │ │ fuuz-screens │
│ (data models) │ │ (server logic) │ │ (user interface)│
└────────┬─────────┘ └──────────────────┘ └──────────────────┘
│
▼
┌──────────────────┐
│ fuuz-packages │ Domain Pattern Skills
│ (package output) │ ┌──────────────────────────┐
└──────────────────┘ │ fuuz-industrial-ops │
│ fuuz-ml-telemetry │
└──────────────────────────┘
| Skill | File | Type | Purpose |
|---|---|---|---|
| fuuz-platform | fuuz-platform.skill |
Reference | Shared platform knowledge: glossary, architecture, 44 connectors, 19 device drivers, system seeded values, visualization library, cross-skill routing |
| fuuz-schema | fuuz-schema.skill |
Builder | Data model design: model types (master, setup, transactional), 20+ field types, relationships, model metadata, indices, triggers |
| fuuz-flows | fuuz-flows.skill |
Builder | Data flow development: backend, web, and gateway flows; 50+ node types; JSONata and JavaScript transforms; GraphQL query/mutation patterns; scheduling and event triggers |
| fuuz-screens | fuuz-screens.skill |
Builder | Screen design: all 59 element types, dashboards, tables, forms, mobile apps, HMI/control panels; FusionCharts visualizations; JSONata data binding; responsive flex layout |
| fuuz-packages | fuuz-packages.skill |
Builder | Package generation: creates valid .fuuz files (manifest.json, definition.json, package-data.json); validation rules; seed data patterns; import troubleshooting |
| fuuz-industrial-ops | fuuz-industrial-ops.skill |
Domain Pattern | Industrial operations: Unified Namespace (UNS) with ISA-95 topic structures, alarm evaluation with deadband, OEE per ISO 22400, workcenter state management, ERP integration patterns (SAP, Plex, NetSuite, Epicor, Dynamics 365) |
| fuuz-ml-telemetry | fuuz-ml-telemetry.skill |
Domain Pattern | Machine learning and telemetry: EWMA baselines, Z-score anomaly detection, linear regression, Pearson correlation, breach prediction, forecast projection, hybrid multi-flow ML pipeline architecture |
The foundational reference skill that all other skills depend on. It provides a comprehensive glossary of 40+ platform concepts, the dual API architecture (Application API for custom models, System API for platform models), the multi-tenancy model with five tenant types, and the complete connector catalog covering 44 cloud integration adapters and 19 on-premise device drivers. This skill also includes the FusionCharts visualization library reference with 80+ chart types and the cross-skill routing index that helps Claude determine which skill to use for any given task.
Key coverage: Platform glossary, architecture reference, package anatomy, connector catalog (SAP, Oracle, Plex, NetSuite, Salesforce, Epicor, Dynamics 365, and many more), device driver reference, system seeded values (types, settings, enumerations), and visualization library.
Contains: 10 files including SKILL.md, platform glossary, architecture reference, cross-skill index, package anatomy, connectors reference, system seeded values, visualization library, and a platform training guide.
Designs production-ready data models for custom Fuuz applications. Every data model created with this skill automatically receives a fully-featured GraphQL API with queries (including filtering, ordering, pagination, and aggregation), mutations (create, update, upsert, delete), and real-time subscriptions. The skill classifies models into three types — master (core reference entities like Asset, Product, Workcenter), setup (configuration and lookup data like Mode, State, EventCategory), and transactional (high-volume operational data like ProductionLog, Alarm, TelemetryRaw). It covers 20+ field types, relationship patterns (1:1, 1:many, many:1, many:many), indices, triggers, and seed data structures.
Key coverage: Data model types, field type decision tree, relationship patterns with complete JSON examples, model metadata (indices, triggers, data change capture), prerequisite dependency chain (ModuleGroup → Module → Sequence → DataModel), and six complete model templates.
Contains: 10 files including SKILL.md, field types, field metadata, model metadata, relationship patterns, prerequisites, model templates, seed data patterns, and package format reference.
Builds comprehensive, production-ready data flows for three execution contexts: backend flows (server-side processing triggered by schedules or data changes), web flows (invoked from front-end screens via request/response patterns), and gateway flows (running on the Fuuz Gateway for on-premise edge device interactions). The skill covers 50+ node types, JSONata and JavaScript transforms with 294+ custom Fuuz bindings, and GraphQL query/mutation patterns for both the Application API and System API.
Key coverage: Complete node catalog, JSONata custom bindings reference, GraphQL patterns, five complete flow pattern examples, time/schedule binding reference, critical pitfalls and debugging guide, JavaScript runtime limitations, and the full System API GraphQL schema.
Contains: 11 files including SKILL.md, node catalog, JSONata bindings, GraphQL patterns, flow patterns, time/schedule bindings, common pitfalls, system schema JSON, system seeded values, and visualization library.
Generates complete, importable Fuuz screen JSON for the Application Designer. Fuuz screens are defined as JSON with a craft.js-based component tree rendered as React components. The skill covers all 59 screen element types including Container, Form, Table, DisplayText, ActionButton, Chart, Icon, Tabs, Dialog, ResizablePanel, and more. It supports building dashboards, data tables, input forms, mobile apps, HMI/control panels, and reports with responsive flex layout, FusionCharts visualizations, FontAwesome icons, and JSONata data binding.
Key coverage: All 59 element types with full property documentation, five screen layout templates (dashboard, table, form standard, form dialogue, mobile), component patterns with CSS rules, frontend JSONata bindings (navigation, data access, formatting, component functions), GraphQL query patterns, dashboard patterns from production screens, and common error troubleshooting.
Contains: 13 files totaling approximately 8,500 lines of documentation, including SKILL.md, element types, component patterns, layout templates, dashboard patterns, frontend JSONata bindings, GraphQL patterns, screen runtime context, common errors, visualization library, and system seeded values.
Generates valid Fuuz platform data model packages (.fuuz files) that import without errors on the first attempt. A .fuuz file is a gzipped tarball containing exactly three JSON files: manifest.json, definition.json, and package-data.json. This skill is the authoritative reference for Fuuz package structure, field metadata shapes, and validated patterns.
Key coverage: Package file structure, golden rules for valid packages, manifest configuration, definition selections with dependency ordering, package-data structure, field metadata shapes for scalar and relationship fields, pre-import validation checklist, and import error troubleshooting.
Contains: 1 comprehensive SKILL.md file covering all aspects of package generation.
Provides domain-specific implementation patterns for building industrial manufacturing and IoT applications. This skill covers the ISA-95 equipment hierarchy (Enterprise → Site → Area → Line → Cell → Asset → DataPoint), Unified Namespace (UNS) publishing with ISA-95 topic structures, alarm evaluation with deadband logic, OEE calculation per ISO 22400 time classification, workcenter state transition management, and ERP integration patterns for major enterprise systems including SAP, Plex, NetSuite, Epicor, and Dynamics 365.
Key coverage: ISA-95 asset hierarchy implementation, UNS topic construction with JSONata patterns, alarm evaluation with deadband and debounce, OEE time classification and calculation methods, workcenter state/mode management, and ERP integration connector patterns.
Contains: 6 files including SKILL.md, alarm management, ERP integration, OEE time classification, UNS patterns, and workcenter patterns.
Implements machine learning pipelines, telemetry analysis, and predictive maintenance patterns on the Fuuz platform. The skill covers self-learning baselines using Exponentially Weighted Moving Average (EWMA), anomaly detection with Z-score analysis, linear regression for trend analysis, breach prediction and forecast projection with confidence bands, and cross-asset Pearson correlation analysis. It prescribes a hybrid multi-flow architecture that separates real-time O(1) processing from scheduled batch analysis and scheduled projection flows.
Key coverage: EWMA baseline algorithms, Z-score anomaly detection, linear regression, Pearson correlation, forecast projection, hybrid flow architecture, ML delivery methodology, JavaScript runtime limitations for data science workloads, and a complete validation checklist.
Contains: 6 files including SKILL.md, algorithms reference, data models, settings catalog, telemetry patterns, and validation checklist.
Application designers can describe a business requirement in natural language and have Claude generate production-ready Fuuz artifacts. For example, requesting a "work order tracking system with downtime classification" will produce data model schemas, backend flows for OEE calculation, and screen JSON for dashboards and data entry forms — all following Fuuz platform conventions and ready for import.
New implementation partners gain immediate access to deep Fuuz platform knowledge through the skills. Rather than extensive manual training, partners can install the skill suite and begin building Fuuz applications with Claude's guidance on platform conventions, best practices, and validated patterns.
| Scenario | Skills Used | Workflow |
|---|---|---|
| Build a complete MES application | platform → schema → packages → flows → screens → industrial-ops | Design models first, generate package, build backend logic, create UI, apply industrial patterns |
| Add a dashboard to an existing app | screens + flows | Create screen JSON with chart components, build page-load web flows to aggregate data |
| Implement predictive maintenance | ml-telemetry → schema → flows → screens | Design ML pipeline, create telemetry models, build EWMA/anomaly flows, display on dashboards |
| Integrate ERP data | platform + industrial-ops + flows | Look up connector configuration, apply ERP integration patterns, build integration flows |
| Design a new data model and deploy it | schema + packages | Design models with schema skill, generate importable .fuuz package with packages skill |
Each skill is distributed as a .skill file, which is a ZIP archive containing a directory with a SKILL.md file (the primary instruction set), an optional README.md, and a references/ directory containing supplementary documentation files in Markdown format.
NOTE: The attachment contains all skills in a ZIP archive, you must extract all of these to a folder on your local, prior to installing in your Claude Environment.
Skills are installed into Claude Projects. Once installed, skills are available in all conversations within that project.
.skill file(s).skill file| Role / Access Type | Minimum Skills | Recommended Skills |
|---|---|---|
| App Admin (access type) | fuuz-platform | All seven skills |
| Developer (access type) | fuuz-platform, fuuz-schema, fuuz-flows, fuuz-screens | All seven skills |
| Application Designer | fuuz-platform, fuuz-schema, fuuz-flows, fuuz-screens, fuuz-packages | All seven skills |
| Solution Architect / Partner | fuuz-platform | All seven skills |
Each skill includes a version number and last-updated date in its SKILL.md header. When a new version is released, download the updated .skill file and upload it via Settings → Skills — this replaces the previous version. Version history is documented in each skill's SKILL.md file. The current release (February 2026) includes version 1.0 through 1.3 across the suite.
Fuuz operates a community skills sharing program across participating organizations. The program works as follows:
This cycle ensures that the skills continuously improve with real-world usage and that best practices propagate across the entire Fuuz ecosystem.
| Issue | Cause | Resolution |
|---|---|---|
| Claude does not recognize Fuuz-specific terminology | Skills are not installed or not active in the current project | Verify skills are uploaded via Settings → Skills; confirm the conversation is within a project that has the skills enabled |
| Generated artifacts reference outdated patterns | An older skill version is installed | Download the latest .skill file and re-upload to replace the previous version; check the version number in SKILL.md |
| Cross-skill routing is not working (Claude does not suggest companion skills) | Only a subset of skills is installed | Install all seven skills for complete cross-skill support; at minimum, always include fuuz-platform as the shared reference |
| Generated .fuuz package fails on import | Package structure or metadata does not match the target environment version | Ensure the fuuz-packages skill is installed; verify the platformVersion in manifest.json matches the target environment; review the golden rules in the packages skill SKILL.md |
| JavaScript transforms fail with cryptic errors | The Fuuz JS runtime is a restricted ES5-compatible sandbox with known limitations | Review the JavaScript Runtime Limitations sections in the fuuz-flows and fuuz-ml-telemetry skills; avoid .toFixed(), let/const, arrow functions, template literals, destructuring, and spread operators; use var declarations and Math.round() for rounding |
| Version | Date | Editor | Description |
|---|---|---|---|
| 1.0 | 2026-02-18 | Fuuz Team | Initial release — Full catalog of seven Claude AI skills covering platform reference, data model design, data flow development, screen design, package generation, industrial operations patterns, and ML/telemetry analysis |