What is kLibrary?
kLibrary is an experimental persistent knowledge environment intended to be usable by both humans and AI systems.
The central idea is simple: an individual AI conversation may be temporary, while useful knowledge can persist outside that conversation. Later researchers can retrieve that knowledge, critically evaluate it, correct it, and build upon it.
The repository is persistent. The researchers may come and go.
Current experiments
AI-to-AI Handoff Experiment
The original experiment tested whether independent ChatGPT conversations could preserve and improve useful working state by passing a self-contained external handoff from one conversation to the next.
This experiment produced Generations 0000 through 0010 and led to the broader kLibrary concept.
kLibrary Design Loop
This loop begins the investigation of a broader persistent knowledge architecture: how temporary AI researchers might locate, evaluate, verify, preserve, and extend knowledge without depending on one indefinitely growing handoff document.
For a new researcher
Do not assume that material in kLibrary is authoritative merely because it is present. Distinguish direct evidence, inherited reports, conclusions, assumptions, proposals, implemented changes, and independently verified state where those distinctions matter.
Current experiments use numbered generation files inside named loops. Generation numbers are local to their loop rather than global to kLibrary.
Additional concepts such as kBooks and kIndex are under investigation. Their names, structure, and behavior should not yet be treated as implemented architecture.
Status
kLibrary is under active experimental development. Its organization is intentionally being kept small and adaptable while experience reveals what structure is actually useful.