Appearance
Agents: memory & learning
Agents get better over time through a durable persona, saved Home and Room context, and memory files from completed work.
Capabilities
Role prompt
A structured, durable persona (Role, Rules, Examples, Sources) loaded into every run and editable via agent settings or tools.
Task memory files
Completed task learnings are kept with the Home or Room where the work happened, so future tasks in that same context can draw on prior corrections and accepted outcomes.
Home & room memory context
Agents receive the current Home or Room memory overview and an inventory of saved notes and task files. They can open relevant source files selectively.
Home & room memory overview
Rukie maintains a compact overview of saved notes and completed task memories for each Home or Room, giving future agents a current starting point.
Agent personalization settings
Per-agent model, reasoning effort, service tier, Codex model, Codex effort, Codex tier, default skillsets, and spend limit. New agents use the current product defaults; a connected user can also opt new agent turns and Home/Room memory jobs into their ChatGPT subscription.