A lightweight, expert-informed GraphRAG component designed for high-stakes medical summarization. TAGRAG (Temporary Auxiliary GraphRAG) enhances retrieval-augmented generation by incorporating targeted, real-time knowledge graphs derived from current, peer-reviewed research. Used in the context of informed consent documents, TAGRAG distinguishes between standard-of-care and study-specific risks—helping clinicians and trial designers generate accurate, plain-language summaries for potential study participants. Rather than maintaining a massive, persistent graph, TAGRAG builds a temporary, context-specific knowledge graph on-the-fly using curated PubMed Central data. It identifies relevant disease-specific publications, extracts structured information, and generates verifiable Q&A outputs for downstream use in consent summaries. By integrating TAGRAG, our system doesn’t just summarize—it informs with evidence. This hybrid approach ensures clarity, accountability, and ethical alignment in one of the most sensitive areas of clinical communication.