We added memory consolidation work around the RAG system. The goal is to avoid unbounded, duplicate, or low-value memory growth.
Memory has to remain inspectable. Consolidation should improve future retrieval without turning memory into invisible prompt stuffing.
A pronoun resolution path was added to reduce failures on follow-ups like 'what about that' and 'what was her role'.
Conversation context handling was added so follow-up questions can reuse prior state instead of being treated as isolated prompts.
We clarified the boundary between conversation memory, retrieved evidence, and answer synthesis in the Agentic RAG runtime.