Elasticsearch Memory MCP is a Model Context Protocol server that provides persistent, intelligent memory by leveraging Elasticsearch for vector embeddings and hierarchical categorization. It organizes memories into five categories: identity, active_context, active_project, technical_knowledge, and archived, with automatic category detection and confidence scoring, plus manual reclassification support. The server features Intelligent Auto-Detection with an accumulative scoring system (0.7-0.95 confidence range) and 23+ keyword patterns for context-aware categorization. A Batch Review System enables bulk processing of uncategorized memories with approve/reject/reclassify workflows, delivering faster moderation. A Backward Compatible Fallback loads v5 uncategorized memories with no data loss during upgrades. Optimized Context Loading prioritizes memories in a hierarchical order to reduce tokens by approximately 60-70% and improves relevance ranking. The memory is persisted with vector embeddings for semantic search, session management with checkpoints, and conversation snapshots. The MCP server can be used via available tools like save_memory, load_initial_context, review_uncategorized_batch, apply_batch_categorization, search_memory, and auto_categorize_memories, and is installable from PyPI. It supports Claude Desktop and Claude Code CLI integration.