Agentic Skills Starter Kit

Everything you need to start building multi-agent tools using Agent Skills — the new open standard that lets anyone extend what AI agents can do, no coding required.

This is the complete training I presented to BYU's AI in Business Society in February 2026, packaged up so you can learn at your own pace. It includes the full slide deck, a working demo you can run yourself, and a real agent skill folder you can explore to see exactly how skills are built.

What's in the Kit

• Training Presentation (agent-skills-training.pptx) — 24 slides covering what skills are, how they work, how to build your own, and how they connect with MCP and sub-agents.
• Demo Data File (demo-sales-data.csv) — A sample dataset with 6 months of sales data across 3 regions and 2 products. Used for the hands-on demo during the training.
• Data Insights Agent Skill (data-insights/) — A complete, working agent skill that transforms CSV data into interactive HTML dashboards with charts, KPI cards, and written insights. This is the demo skill from the presentation, and it's also a great reference for how a real skill is structured.

Getting Started

Step 1: Watch the Presentation

Open agent-skills-training.pptx and read through the slides. The deck is designed to be self-explanatory — it covers the core concepts, shows you the built-in skills that come with Claude, and walks through how to create your own. Takes about 15–20 minutes to read through.

Step 2: Try the Demo

This is the fun part. Open Claude (Desktop or claude.ai with file creation enabled) and try the demo from Slide 10:

1. Upload or paste the contents of demo-sales-data.csv
2. Ask Claude to analyze the data and create a dashboard or detailed spreadsheet
3. Watch the agent skill activate and produce a polished output

You can experiment with different prompts — ask for charts, forecasts, regional breakdowns, or an executive summary. The built-in skills handle it from there.

Step 3: Explore the Skill Folder

This is where you learn how the sausage is made. Open the data-insights/ folder and look around:

data-insights/
├── SKILL.md                     ← Start here. The brain of the skill.
├── scripts/
│   └── clean_data.py            ← Data cleaning utility
├── templates/
│   └── dashboard-template.html  ← HTML dashboard template
├── assets/
│   └── color-palette.json       ← Chart color definitions
└── docs/
    └── visualization-guide.md   ← Chart selection reference

SKILL.md is the file that matters most. It contains the skill's name, a description that tells the agent when to use it, and step-by-step instructions for how to use it. Read it carefully — this is the pattern you'll follow when you create your own skills.

The rest of the folder contains supporting files that the skill references: a Python script for cleaning messy data, an HTML template for dashboards, a color palette for consistent chart styling, and a guide for choosing the right visualization. These show how a skill can be modular — the instructions in SKILL.md point to these files as needed.

Step 4: Build Your Own

Once you've seen how the data-insights skill is organized, you have everything you need to create your own. The presentation covers this in detail (Slides 11–15), but the short version is:

1. Create a folder with a descriptive name
2. Add a SKILL.md file with a name, description, and instructions
3. Add any supporting files (templates, scripts, examples)
4. Drop the folder into your skills directory

That's it. Your agent discovers it automatically on the next conversation.

Questions?

If you build something with this kit or have questions, reach out to me on LinkedIn — I'd love to hear about it.