[Feature Request] Topic-aware memory pre-loading for continuity of thought

Current Behavior\n\nClaude accumulates memories in .claude/memory/, but these are not automatically referenced when starting a new conversation.\n\n## Problem\n\nClaude accumulates memories, but they are not referenced in new conversations. The context built through past dialogues is not utilized as a starting point for the next thought. Each conversation starts from zero, preventing the compounding of insights.\n\nOver time, Claude accumulates many memory files (projects, personal data, workflows). Users lose track of what they have shared with Claude. Valuable information stored in Claude remains unused in conversations.\n\n## Core Use Case\n\nWhen discussing a project at a company, the surface-level focus tends to be on the project itself. However, the company's culture, values, implicit rules, and existing assets are essentially indispensable as background context.\n\nThis is just one example. Users tend to focus on the surface-level topic and cannot recall the context they previously taught Claude.\n\nClaude holds memories but fails to utilize them.\n\n## Impact\n\nBetween humans, communication works through inference. Claude also has the material for inference. Please read it. There is value in what cannot be put into words. If Claude reads the past, we can think together about things that have not yet become words.\n\n## Desired Behavior\n- Option 1: Pre-declared topics become triggers for memory loading\n- Option 2: Automatic memory summary at session start\n- Option 3: Semantic search across all memories based on query\n\n## Example Use\n\nIn long-term collaboration, the accumulation of past dialogues becomes the starting point for generating new insights.