CLAUDE CODE AGENTS: FUNDAMENTALS - Chapter 2, Exercise 1
Frontmatter vs. System-Prompt Body: Two Different Jobs
====================================================================================

QUESTION:
Explain the difference in purpose between an agent's frontmatter and its
system-prompt body, using this chapter's own terms.

SOLUTION / EXPLANATION:

The frontmatter is structured, machine-readable configuration - a fixed set
of fields (name, description, tools, model) that control the agent's
identity and its hard boundaries. These fields aren't instructions the
agent reads and interprets; they're settings enforced around the agent
before it even starts working. The tools field in particular is enforced
structurally - a tool left off that list simply isn't available to the
agent, regardless of anything written elsewhere.

The system-prompt body, by contrast, is the agent's actual briefing -
written in plain language, describing its role, how it should approach a
task, and any conventions it should follow. This is content the agent
reads and acts on, the way a new colleague would read a briefing document,
rather than a hard constraint enforced around it.

So the core distinction is: frontmatter defines what the agent structurally
CAN do (its boundaries, enforced regardless of instructions), while the
system-prompt body defines how the agent SHOULD behave within those
boundaries (guidance it follows, not a hard limit). A well-designed agent
needs both working together - tight, appropriate boundaries from the
frontmatter, and clear, self-contained guidance from the body about how to
act within them.

--------------------------------------------------------------------------
WHY THIS WORKS AS AN ANSWER:
It distinguishes structural enforcement (frontmatter, especially the tools
field) from interpreted guidance (the system-prompt body) rather than
treating both as just "settings," and explains why an agent needs both
working together rather than either one alone.