Spore Core

Spore Core is about treating an agent harness as something you build, not a large object you inherit and hope behaves.

Why I built it

The interesting part of an agent system is often the loop: what plans, what acts, what observes, and what gets to change its mind. That loop should be a choice, not a welded-on personality trait.

The idea / architecture

Spore Core frames the harness as a composition point. Models, tools, context, hooks, and events can meet there, while the loop strategy stays pluggable. A simple loop can remain simple; a more involved one can add steps without requiring the whole harness to become a different project.

Put another way: the loop is not the harness. It is one of the things the harness is allowed to compose.

What it became / current state

The source notes currently establish the central idea and its visual language: a composable harness with replaceable pieces and loop strategies. The implementation history and present release state still need to be filled in from the project inventory.

Notable details

• The reference loop is deliberately small: plan, act, observe.
• The visual can expand that loop into plan, act, validate, reflect, observe—without implying that every Spore Core setup must use those steps.
• Motion is intended to remain subtle, work without JavaScript, and respect reduced-motion preferences.