FrameworX provides practical AI capabilities for industrial automation.
Platform Overview | AI Integration, from AI-assisted solution building to runtime data queries, machine learning deployment, and portable AI skills to Claude and other AI agents.
Why AI Integration Matters
Engineers using AI copilot tools with FrameworX report productivity improvements of 2× to 10× for configuration tasks, depending on complexity. This isn't a marginal improvement; it's a fundamental change in how industrial applications are built.
FrameworX provides practical AI integration across three four areas: design-time assistance, runtime queries, and machine learning deployment, and portable AI skills for Claude and other AI agents.
The Next Frontier: A decade ago, the industry focused on OT-IT integration. Now the frontier is OT-AI integration. Similar The same architectural decisions that enabled FrameworX OT-IT : integration — consistent namespaces, managed code, open interfaces; now , now enable native AI integration.
Table of Contents maxLevel 2 minLevel 2 indent 10px exclude matters style none
AI Capabilities Overview
| Capability | What It Does | Use For | Unique Features |
|---|
| MCP for Designer | AI models help build solutions | Faster configuration, bulk object creation | AI Tools (Claude, Copilot) directly manage the Designer UI. |
| 16 tools for creating all solution elements with visual feedback. | |||
| MCP for Runtime | AI models query live solution data | Dashboards, troubleshooting, automated reports | AI tools (Claude, Copilot, cursor) directly access runtime data. Visual feedback: orange border and badge when AI is connected. |
| FrameworX Skills Library | Playbook patterns for complex application builds | Alarm pipelines, edge ML, MQTT integration | Comprehensive build playbooks that MCP tools use to guide solution construction for specific industrial patterns. |
| ML |
| Integration | Deploy trained ML models | Anomaly detection, predictive maintenance | Use directly ML.NET 4.0 in C# |
| Use directly all Python 3.7+ libraries | ||
| Ask FrameworX Docs | ChatGPT trained on documentation | Get help, ask questions. |
| Sidebar shortcut: AI Ask |
| Docs |
| Leverages OpenAI tools | |||
| Claude Product Knowlege Skill | Gives Claude deep product knowledge — architecture, syntax, protocols, modules, and 200+ documentation links | Answer FrameworX questions accurately, guide configuration, link to the right docs page | Open standard (agentskills.io). Works in Claude.ai, Claude Desktop, Claude Code, GitHub Copilot, Cursor. The Claude equivalent of Ask FrameworX Docs. |
| Info | ||
|---|---|---|
| ||
MCP for Designer is a game changer on for how to develop solutions are developed. If that is new for you, go this bries summary page: MCP For Designer In Action |
MCP for Runtime: Query Live Data
Connect AI models like Claude to your running solution.
What AI can do:
- Read tag values from the UNS
- Query alarms and historian
- Create custom queries via Script Classes
Understanding the Two Kinds of "Skills"
FrameworX uses the word "skill" in two distinct contexts. Understanding the difference helps you get the most from AI integration.
| Claude Skill | FrameworX Skills Library | |
|---|---|---|
| What it is | A portable SKILL.md file you install in Claude (or other AI agents) | Solution-building playbooks inside FrameworX that guide MCP tools through complex patterns |
| Where it lives | In your AI agent (Claude Desktop, Claude Code, GitHub Copilot, Cursor) | In the FrameworX documentation, accessed by MCP tools via search_docs |
| What it does | Gives Claude comprehensive product knowledge — architecture, syntax rules, all table types, 100+ protocol connectors, common questions, and 200+ verified documentation links. The Claude equivalent of Ask FrameworX Docs on ChatGPT. | Provides detailed step-by-step build instructions for specific industrial patterns (alarm pipelines, edge ML, MQTT integration). |
| Who uses it | Anyone asking Claude about FrameworX — evaluators, engineers, integrators, even without Designer installed | MCP tools during active solution building — requires Designer and an open solution |
| Standard | Agent Skills open standard (works across AI platforms) | FrameworX-specific (accessed through MCP tool search_docs(label='skill')) |
In short: The Claude Skill gives Claude product knowledge for answering questions and guiding decisions. The FrameworX Skills Library gives MCP tools build playbooks for constructing complex solutions. They serve different purposes and work independently.→ See AI MCP for Runtime Connector
MCP for Designer: AI-Assisted Configuration
Let AI help you build solutions faster with 17 specialized tools.
What AI can do:
- Create Tags, Displays, Alarms, Devices, and all solution objects
- Generate Canvas and Dashboard displays with proper patterns
- Search protocols by equipment name (Siemens, Allen-Bradley, Modbus)
- Track changes, cross-references, and audit trails
- Access documentation and download example solutions
Key Features:
- Visual Indicator — Orange border and "AI MCP" badge show when AI is connected
- MCP Category System — Tracks AI-created vs manually-created objects
- Wizard Symbols — Pre-built industrial symbols (BLOWER, MOTOR, PUMP, TANK, VALVE)
- Protocol Search — Fuzzy matching finds protocols by vendor nameVisual Feedback — Designer UI shows changes as they happen
AI Knowledge of FrameworX
When using MCP for Designer tools, it automatically feeds the AI chat session with a summary of the platform architecture — covering the unified model, tool usage patterns, build recipes, and behavioral principles. This means Claude reaches expert-level understanding of FrameworX immediately, without any manual setup.
When working with chat sessions and AI models not using our MCP for Designer tools, the Claude Skill provides comprehensive product knowledge (architecture, syntax, protocols, modules, and documentation links), and the Platform Architecture Reference page can be fed to any AI for deeper context.
→ See MCP for Designer In Action
→ See AI MCP for Designer Connector
FrameworX Skills Library
The FrameworX Skills Library contains comprehensive playbooks for complex application patterns. These are not files you install in Claude — they are documentation pages that the MCP tools access automatically via search_docs(label='skill') when building solutions that require specific industrial patterns.
Available skills:
- — End-to-end alarm configuration
- — Machine learning at the edge
- — MQTT and SparkplugB integration
→ See for creating your own FrameworX skills
MCP for Runtime: Query Live Data
Connect AI models like Claude to your running solution.
What AI can do:
- Read tag values from the UNS
- Query alarms and historian
- Create custom queries via Script Classes
→ See AI MCP for Runtime Connector
ML.NET Integration: Deterministic Machine Learning
Deploy trained models for consistent, repeatable results.
Use cases:
- Anomaly detection
- Predictive maintenance
- Quality prediction
In addition to ML in C#, the Python 3 native integration allows the usage of all Python analytics libraries.
→ See AI ML Integration Connector
→ See Python .NET Integration
Ask FrameworX Docs
ChatGPTin getting ChatGPT trained on FrameworX documentation.
You can click Click here → AI Ask Docs, or use the shortcut on the left top sidebar.
Reference Link
sidebar.
Claude Skill: Portable AI Product Knowledge
The FrameworX Claude Skill gives Claude (and other AI agents that support the Agent Skills open standard) comprehensive knowledge of the FrameworX platform. It is the Claude equivalent of Ask FrameworX Docs on ChatGPT.
What the skill covers:
- Product architecture — UNS, Four Pillars, server/client runtime, built-in namespaces
- Syntax reference — tag paths, slash vs dot rules, @ prefix, UserType patterns, common mistakes
- All module configuration — table types, key fields, and relationships for Devices, Alarms, Historian, Displays, Datasets, Scripts, Reports, Security
- 100+ protocol connectors — organized by vendor, standard, and industry with configuration patterns
- Common questions — verified answers to frequently asked questions
- Industry use cases — 38 documented deployments across manufacturing, oil & gas, water, power, food & bev, transportation
- 200+ documentation links — verified URLs so Claude links to the right docs.tatsoft.com page every time
The skill and MCP tools are independent. Use the skill alone for product knowledge and documentation guidance. Add MCP when you are ready to build solutions hands-on. They do not require each other.
→ See for installation instructions and download
Reference
→ For the technologies → For the technical architecture that enables this, see AI-Ready by Design
In this section...
| Page Tree | ||||
|---|---|---|---|---|
|