Problem
Useful context is distributed across repositories, branches, documents, memory, handovers, runtime evidence and historical/current sources. Loading more context does not automatically produce better orientation.
SRCB
SRCB builds, enriches, reconciles and projects a verified context model of a bounded subject from distributed evidence before and during substantial AI-assisted work.
Core question
What must this consumer understand about this bounded SUBJECT now, what supports it, what conflicts, and what remains unknown?
Protections
context acquisition != truth by accumulation
retrieval != authority
memory != current evidence
projection != source truth
Composition
SRCB composes existing capabilities rather than inventing a new memory system: ORBIT for orientation, SORR for semantic resolution, CARD/POINTER for representation and routing, and target-specific sources for claim authority.
Scope
SRCB governs bounded context acquisition and projection. It keeps retrieved material, remembered material and source authority distinct, and focuses the resulting context on the consumer and subject rather than accumulating everything available.