Introduction to Agent Skills cover
← Back to Courses

Introduction to Agent Skills

Advanced

Reusable instructions Claude applies automatically. Master SKILL.md, triggering, access control and enterprise distribution — and stop repeating yourself.

1 Day / 7 HoursAI & Generative AILive online with recordings · On-site · Private cohort
HRD Corp Claimable
HRD Corp Claimable
TP No · 10001731836

Program Fee

$1,200

1-on-1 doubt clearing · WhatsApp group · Post-training Q&A

Request Syllabus / Call Back

Program Overview

If you find yourself pasting the same instruction into Claude every day, you should have written a skill. This day teaches how. You install a skill, watch it trigger, and — more instructively — watch it fail to trigger, because the description is the single hardest and most important line you will write. You learn the SKILL.md format, progressive disclosure so only what the task needs gets loaded, and the distinction between a skill, a prompt and an MCP server. The core module is design and testing: scoping one skill to one job, building a trigger test set with positive and negative cases, and tuning the description against measured accuracy rather than vibes. You finish with access control, versioning and deployment to a shared team repository.

What You'll Learn

  • Explain what a skill is, and how it differs from a prompt and from an MCP server
  • Describe how triggering works, and why the description does all the work
  • Apply progressive disclosure so only what is needed gets loaded
  • Write a correct SKILL.md with frontmatter, body and bundled resources
  • Write a description that triggers reliably
  • Scope one skill to one job
  • Build a trigger test set with positive and negative cases
  • Tune a description against measured accuracy rather than intuition
  • Diagnose a skill that never fires, and one that fires constantly
  • Restrict what a skill may do
  • Version, review and distribute skills across a team

Who Can Join

  • Developers repeating the same instruction daily
  • AI and platform engineers
  • Technical leads standardising a team’s AI workflow
  • Anyone maintaining a growing pile of copy-paste prompts

PREREQUISITES

Experience using Claude Code or the Claude API

Comfortable with Markdown and version control

Curriculum

The idea

Skills, prompts and MCP

How triggering works

Progressive disclosure

Lab 0

Total: 7h across 4 modules

Tools & Technologies

Agent SkillsSKILL.mdClaude CodeClaude DesktopGitprogressive disclosure

Projects Included

  • Lab 0 — A skill installed, triggering observed, and made to fail
  • Lab M01 — A SKILL.md for one of your team’s recurring conventions
  • Lab M02 — A trigger test set, and a description tuned against it
  • Lab M03 — A skill packaged with access control and deployed to a shared repository

Official Certifications

Course Completion Certificate — issued by CertTulen Academy.
Course Completion Certificate — issued by Anthropic.
Exam — Not applicable.

Assessment is continuous and practical: 4 Applied Learning Labs (60%) and an end-of-course practical assessment (40%). Passing mark 80%, plus completion of all 4 labs and full attendance. The CertTulen Course Completion Certificate is awarded on meeting that standard.

Questions? Contact us for a free counseling session.

Have questions about this course?

Send us your details — we'll respond within 24 hours.

Chat with us