DRY

DRY ("Don't Repeat Yourself") is a classic software engineering principle holding that logic or behavior should be defined once and reused rather than duplicated. In the context of AI agent engineering, it is applied to the design of skills and resolver tables to keep agent architectures clean and non-redundant.

Application to Agent Skill Design

In the Inside YC's AI Playbook talk, DRY is invoked as a guiding criterion when auditing the skill inventory that feeds a resolver. The recommendation is to survey all existing skills and ask whether any are duplicative — if two skills overlap in purpose, they should be collapsed into a single parameterized skill rather than maintained as separate entries. This keeps the resolver table unambiguous and prevents the agent from having to arbitrate between near-identical options. (16:41)

DRY is paired with the MECE principle (Mutually Exclusive, Collectively Exhaustive) as a dual standard: skills should neither overlap (DRY) nor leave gaps (MECE), so that the resolver can route cleanly to exactly one appropriate skill for any given task. (16:41)

Summary

Principle	What it guards against
DRY	Redundant, overlapping skills
MECE	Missing coverage or ambiguous routing

Together they define a quality bar for the resolver's skill table: one well-parameterized skill is strictly preferable to ten narrowly scoped, partially overlapping ones.