Skip to main content

Agent Skills Guides & Articles

Agent skills package reusable guidance for tasks that need domain knowledge. The central question is not how much expertise a skill contains, but whether an agent performs better with it than without it.

Start with the lessons from six months of skill development, then read the regression analysis before expanding your own guidance. The prompting article explores how separating responsibilities can help. Together, these articles offer a practical path from writing instructions to evaluating their effect.

By Sajeetharan Sinnathurai, Principal Product Manager at Microsoft.

Start here

  1. What Six Months of Building AI Agent Skills Taught Me About Writing for Machines

    Lessons from six months building AI coding agent skills: focused domain guidance, realistic evaluation, and knowing when more instructions make outcomes worse.

  2. The Regression Tax: Why Your Agent Skills Might Be Breaking More Than They Fix

    Are agent skills always worth using? A new paper reveals that regressions offset 59% of gross gains — and the best skills win by breaking less, not helping more.

  3. Smart Prompting, Stronger AI Agent Outcomes — What Actually Works

    Over the past few months, one prompting pattern has consistently produced better AI agents for me. Not bigger prompts. Not more skills. Just a better separation of responsibilities.

All Agent Skills articles

Newest first. Publication dates distinguish recent guides from historical walkthroughs.

Search all blog posts | Complete archive | Subscribe via RSS

Ask Sajee

A contact guide, not live chat.

What would you like help with?