Video course · 9 chapters · 31 min · certificate
Build AI Agent Skills That Actually Work
Build agent skills that reliably automate work: what skills are, which types to build, the right mindset and methods, seven best practices, and how to test and optimize them.
What you'll learn
- Explain what an agent skill is
- Distinguish general and personal skills
- Decide when a task deserves a skill
- Build skills from real runs, not one-shot prompts
- Apply seven structural best practices
- Test and iterate skills until reliable
Chapters
9 chapters · 30:56-
1:02
01Intro Members
Why skills matter
Agents in Claude Code, Cowork and Codex are powerful, but reliable automation depends on great skills.
-
1:02
02Basics Members
What skills are
A skill is a preset instruction for a specific task, packaged as a SKILL.md with optional files.
-
2:02
03Types Members
Types of skills
General skills anyone can download versus personal skills built for your own workflows.
-
3:20
04Mindset Members
The skill-building mindset
When to build a skill: tasks you repeat, with a known good process, like onboarding a new hire.
-
4:30
05Methods Members
Skill-building methods
One-shot prompting rarely produces good skills; build from a real, successful run instead.
-
10:55
06Practice Members
Seven best practices
Structuring SKILL.md well: clear triggers, steps, examples, references and constraints.
-
1:01
07Tools Members
A skill builder skill
Using a skill-creator skill to capture a chat process or optimize existing skills automatically.
-
3:37
08Testing Members
Testing skills
Most people skip testing; run skills on real inputs and check the outputs.
-
3:27
09Iteration Members
Optimizing skills
Improving skills over several iterations: don't expect perfection, use them, and refine from real results.
Jobs this course suits
Our AI checked this course against 500 jobs; these get the most out of it. Each job links to its learning path.