Lessons 1-2 Free Beginner

Give Your AI Agent Tools and Skills

Teach your AI agent a skill and connect it to one app in ChatGPT, Claude or Copilot, with every write set to ask first. 8 lessons, first 2 lessons free.

8 lessons
2 hr 9 min
Certificate Included

Your agent answers from its instructions and files when you paste in the work. This course gives it two new things: a skill, which is one of your procedures written as a file it loads when a request matches, and an app connection, so it can read where your work already lives. You build both in the AI tool you already use: ChatGPT, Claude or Microsoft Copilot (formerly Microsoft 365 Copilot) with a work or school account.

You start by finding what your account has already connected, and removing what your job doesn’t need. Then you turn one procedure from your Agent Card into a SKILL.md file, install it, and watch it fire on a request it covers. You test its description with requests that should and shouldn’t use it, connect one app with reads allowed and every write set to ask you first, rate every tool’s risk, add a built-in tool, and learn to vet a connector or plugin before you install it.

Every product step follows the vendors’ own help pages, read on 13 and 14 November 2026, with a way around anything your plan lacks. You finish with version 2 of your agent: one skill you keep as a file, one connected app, a tool risk table and an updated Agent Card. Lessons 1 and 2 are free.

What You'll Learn

  • Identify the apps or connectors, skills and plugins in your AI tool, where each is managed, and what each adds to your agent
  • Apply one procedure from your Agent Card as a skill (a SKILL.md file) and check that your agent uses it
  • Evaluate a skill's description by testing requests that should and should not trigger it
  • Use per-tool permissions to connect one app with reads allowed and every write set to ask first
  • Use a built-in tool such as web search or file creation in one step of your agent's job, and explain its limits
  • Analyze each tool's risk (read or write, reversible or not, the data it reaches, whose account) and evaluate a connector or plugin before you install it

After This Course, You Can

→ Turn a procedure your team repeats into a skill file that works in ChatGPT, Claude or Copilot
→ Connect an AI agent to a work app so it reads what it needs and asks you before it changes anything
→ Write skill descriptions and test them, so your agent uses a procedure when it should and leaves it alone when it shouldn't
→ Decide which tools an agent needs, and which it should never have, before anyone else relies on it
→ Show a skill, its trigger test and a tool risk table in a review or an interview, as evidence you can extend AI agents safely

What You'll Build

Portable Skill Folder
A SKILL.md skill made from one of your own procedures, with a tested description and a record of which requests it fires on.
Tool Risk Table
Every tool your agent can use, rated by read or write, reversibility, the data it reaches and whose account it uses, with each write set to ask or blocked.
Give Your AI Agent Tools and Skills Certificate
A verifiable credential proving you can teach an AI agent a reusable skill, connect it to your apps with least privilege, and vet the tools you add.

Course Syllabus

8 lessons · 2 hr 9 min

Prerequisites

  • Version 1 of your agent from the course Build an AI Agent in ChatGPT, Claude or Copilot, or any project or agent with instructions you have tested
  • An account in ChatGPT or Claude, or in Copilot with a work or school Microsoft account
  • One app account your agent's job uses, such as a calendar or a shared drive (Lesson 4 connects it)

Who Is This For?

  • Owners of an agent built in the previous course, ready to give it more to do
  • Anyone with a ChatGPT or Claude project, or a Copilot agent, who repeats the same procedure in it
  • Team leads who want one written procedure their team's AI tools can all follow
  • Professionals who want an agent to read their apps without letting it change anything unasked
The research says
56%
higher wages for professionals with AI skills
PwC 2025 AI Jobs Barometer
83%
of growing businesses have adopted AI
Salesforce SMB Survey
$3.50
return for every $1 invested in AI
Vena Solutions / Industry data
We deliver
440+
Courses
Teachers, nurses, accountants, and more
2
free lessons per course to try before you commit
Free account to start
9
languages with verifiable certificates
EN, DE, ES, FR, JA, KO, PT, VI, IT
Start Learning Now

Frequently Asked Questions

Do I need to know how to code?

No. A skill is a plain-text file with a name, a description and numbered steps, and the lessons give you a prompt that drafts it from your Agent Card. Connecting an app happens in your tool's settings, and every exercise is a prompt you paste into a chat.

Which AI tools does it cover, and which plan do I need?

ChatGPT, Claude and Microsoft Copilot. The vendors' own pages disagree on which plans get skills: OpenAI's pricing page lists Skills on Free, Go, Plus and Pro while its help page lists business plans, and Anthropic's help centre lists Skills on the Free plan while its docs list paid plans only. Anthropic's pricing page lists connectors on Free, Pro, Max and Ultra. For Copilot you need a work or school Microsoft account; Agent Builder can read your apps as knowledge where your organisation and licence allow it, but can't be given actions in them. Every lesson gives a way around a feature your plan lacks.

Will a skill I make in one tool work in another?

The file will. A skill is a folder with a SKILL.md file in the Agent Skills format, an open standard: Anthropic says more than 70 agent platforms, ChatGPT among them, have integrated it, and Microsoft says Copilot Chat supports it. The upload differs: Claude wants the folder inside the ZIP, while Copilot wants SKILL.md at the top level of the ZIP (Copilot Chat also takes the file on its own). Lesson 2 shows both.

Will my agent change things in my apps without asking?

Not with the settings this course teaches. You connect one app, let it read, and set every action that creates, changes, sends or deletes to ask you first. OpenAI's help page says its Allow all actions setting carries elevated risk, and the course never turns it on. Microsoft's documentation says Agent Builder doesn't support actions in other services: it reads your apps as knowledge.

Do I have to build an MCP server or a plugin?

No. You connect an app from your tool's own directory. Lesson 7 shows when a custom connector, an MCP server someone runs, or a plugin is needed and what each tool's plans allow, then gives six questions to ask before you install one. Making your own plugin is optional.

Is this course free?

Lessons 1 and 2 are free with a free account. The rest of the course and the certificate come with Pro.

Will I get a certificate?

Yes, with Pro: finish every lesson and score at least 70% on the quizzes for a verifiable certificate.

How long does it take?

Eight lessons; the course page shows the total time. Each lesson ends with one piece of your agent's second version.

2 Lessons Free