Give the operator somewhere to start.

A blank workspace makes every setup an improvised collection of prompts. Identity, tool configuration and historical notes become tangled. When the agent learns something important, there may be no durable place to put it or a consistent structure for the next session to read.

We built it for people assembling an AI operator who would otherwise copy a pile of configuration files by hand. It suits teams exploring repeatable agent setups as well as operators who need a clear place for identity, everyday context and accumulated learning.

We built a published plugin that creates identity and working-rule files, a memory index, daily logs and a starter learned rule. Optional components add scheduled briefings and four roles: orchestrator, coder, security and researcher. The configuration lets an operator choose which parts to enable.

Three registered tools inspect scaffold status, create an instinct and create or update an evergreen reference. Existing files are not overwritten. Once the workspace exists, its identity and context belong to the operator rather than being reset by the next installation.

On first load, the kit checks for missing files and fills those gaps from templates. Daily records capture events; references retain lasting knowledge. Instincts record a trigger, action, rationale and evidence with a confidence score. The kit provides the structure and tools; the operator still has to populate and maintain useful content.

We kept scaffolding separate from the underlying agent runtime and conversation-memory engine. Creating folders is not proof that an operator has learned anything, and an optional role configuration does not guarantee good coordination. The kit makes a disciplined starting point reproducible while leaving those responsibilities explicit.