A place for context to live
Every new conversation shouldn't require telling the whole story again. The Library is being built to preserve meaningful history, decisions, source records, and unfinished work — alongside a rolling view of recent activity. The goal is simple: when an intelligence arrives, it can understand what's been happening and continue with context instead of starting over.
That context is meant to belong to its owner, not to a particular model provider. GML is designed to carry it across models and tools, with explicit boundaries for what may be read, shared, or acted upon. Private material stays private unless its owner chooses to share it.
GML is under active development. This is the direction we're building and testing, not a public service available today.