Front Matter & Progressive Disclosure

The differences between skills and agents.md are actually very small. The skill basically just gets called by the agent as and when it needs it instead of always being added to the system prompt. A couple of terms worth mentioning are progressive disclosure and frontmatter.

Frontmatter is effectively like metadata. If you were writing a blog post and you’ve got your title and your tagline, you’re writing the metadata, the SEO data, to say, this is what a search term should hit. When the user types these words, it should hit the tagline and description. So if the post is about AI, you would have AI in there somewhere.

Frontmatter is that for these documents. It looks like this:

name: Newsletter-content-writer
description: Use this skill when asked to write any newsletter content.

The agent itself prepopulates all of those titles and descriptions. It can look at that and say, “Okay, I’ve taken in the system prompt. That’s the first thing that it knows, and then it’s also taken in the agents.md file to know any instructions the user has specified. And then it can also see there are ten skills available. So it now knows that if you ask anything about newsletter content, it will pick up those skills and use them proactively.”

And that leads to what progressive disclosure is. It just means the agent loads skills itself as needed. You don’t have to specify, “Use this skill to do this task.” Although, it’s common to do that anyway just to make sure the agent is definitely using the right skill.