NightFeats @ MMU-RAGent NeurIPS 2025: A Context-Optimized Multi-Agent RAG System for the Text-to-Text Track

NightFeats is a multi-agent retrieval-augmented generation (RAG) system introduced at NeurIPS 2025, which achieved Best Dynamic Evaluation in the text-to-text track. The architecture features a structured pipeline with three phases: retrieval, curation, and composition, utilizing techniques such as temporal-semantic reranking and citation-preserving composition. Its performance outperformed proprietary models like Claude-SonnetV2 and Nova-Pro, highlighting the importance of architectural transparency and verifiable evidence in aligning AI outputs with human preferences, which is crucial for practitioners focused on developing robust LLM applications.