Top priorities in rank order, active goals and decisions, the caller’s responsibilities and focus, and the working principles that shape how AI should behave here. Sized to a token budget (about 1,000–3,000 by default) so it can be fetched at the start of every conversation without crowding out the conversation itself. Returns the organization and personal revision numbers, which change whenever the underlying context does.
Parameters
200–8000. Sections are trimmed in priority order when the budget is tight, so the most important context survives.