FlexA-Mem Provides Budget-Adaptive Memory for LLM Agents

FlexA-Mem implements a general memory system that dynamically renders a fresh memory for every agent call by combining a compact timeline with query-specific details. This approach optimizes context window usage by choosing between raw text, excerpts, or verbatim pieces based on remaining token budget.