AI targeted documentation to enhance models for specific tasks

Start | Platform | Modules | Tutorials | How-to | Reference | Connectors | CasesResources


AI Skills Library

Skills are AI playbooks — focused markdown documents that teach the FrameworX AI Agent how to build specific solutions. Unlike documentation pages written for humans, skills are written for Claude and contain step-by-step MCP tool calls, JSON payloads, and verification commands.Start | Platform | Modules | Tutorials | How-to | Reference | Connectors | Skills | Cases | Resources


What Are Skills?

When you ask the AI assistant to build something in FrameworX — "set up alarms," "connect to MQTT," "add anomaly detection" — the assistant searches for a matching skill and follows its instructions. Think of skills as recipes: each one describes exactly what to create, in what order, with what settings, and what to check afterward.

Skills are different from every other type of documentation on this site:

Page Type

Written For

Purpose

Example

Tutorial

Humans

Learn by doing step-by-step

"Your First FrameworX Solution"

How-to Guide

Humans

Solve a specific problem

"How to Configure Modbus TCP"

Reference

Humans + AI

Look up properties and syntax

"AlarmsItems Table Reference"

Solution Example

Humans + AI

Complete working solution to study

"BottlingLine ML Demo"

Skill

AI assistant

Step-by-step playbook the AI follows autonomously

"Alarm Pipeline — Groups, Areas, and Items"

The key difference: you don't read skills yourself (though you can). The AI reads them, then executes the steps on your behalf using MCP tools like write_objects, list_protocols, and get_table_schema.

How it works

  1. You ask: "Add alarms to my solution"
  2. AI searches: search_docs('alarm', labels='skill')
  3. AI finds: Alarm Pipeline skill
  4. AI reads the full skill content
  5. AI executes each step — creating groups, areas, items
  6. AI verifies the result

Quick links


Find Skills


Official Skills (by Tatsoft)

These skills ship with the product and are maintained by Tatsoft. They cover the most common automation scenarios.

Skill

What It Does

Difficulty

First Solution — Tags, Simulator, and Dashboard

Build a working solution from scratch in 5 minutes

Beginner

Alarm Pipeline — Groups, Areas, and Items

Complete alarm system: groups, areas, items, verification

Intermediate

Edge ML Pipeline — ML.NET Anomaly Detection

Sensor tags → ML model → prediction tags → alarms → dashboard

Advanced

MQTT TagProvider Integration

Connect to MQTT broker, auto-discover topics, build asset-navigation UI

Advanced

User & Community Skills

Integrators and end-users can create custom skills that enforce company standards, project conventions, or industry-specific patterns.

Skill

Author

Purpose

Acme Corp — Tag Naming and Structure Standards

Acme Corp Engineering

Company-specific naming, hierarchy, and unit standards

Want to create your own? See the Skill Authoring Guide for the file format, testing steps, and publishing instructions. Or duplicate the Skill Template Reference page and start editing.


Three Tiers — How Skills Are Discovered

The AI searches for skills in three locations, in priority order. Higher tiers override lower tiers if a skill with the same name exists in multiple places.

Tier

Source

Location

Who Creates Them

1 — User

Your machine

ProductPaths.Skills/skill-*.md

You or your integrator

2 — Cached

Downloaded from Tatsoft

AIKnowledge/skills/skill-*.md

Tatsoft (auto-updated)

3 — Online

This documentation site

docs.tatsoft.com (pages labeled skill)

Tatsoft

Override rule: If you create a skill-alarm-pipeline.md in Tier 1, it replaces the official Tatsoft version. This lets you customize any skill without modifying the original.

Tier 3 works today — the AI searches this Confluence site for pages with the skill label. Tiers 1 and 2 require FrameworX 11.0+ with the local skills system enabled.


Explore All Skills


In this section...