LycheeMem

Overview

LycheeMemory is a compact memory framework for LLM agents that provides efficient conversational memory through structured organization, lightweight consolidation, and adaptive retrieval. It implements MCP support as an HTTP server, making it compatible with any MCP client. The framework uses LanceDB for vector storage and supports multiple reranking models to improve memory retrieval accuracy. The project is actively maintained with recent updates including transformer reranker v0 and visual memory modules.

Try asking AI

After installing, here are 3 things you can ask your AI assistant:

• lychee_memory_smart_search

  Default long-term memory retrieval entry point

  Try:Search for information about previous conversationsTry:Find relevant memories about topic X
• lychee_memory_search

  General memory search function

  Try:Search for specific memoriesTry:Find all memories related to Y
• lychee_memory_append_turn

  Add a new turn to the memory

  Try:Append this user message to memoryTry:Save this assistant response to the conversation
• lychee_memory_synthesize

  Synthesize information from memory

  Try:Create a summary of recent conversationsTry:Generate insights from my memory
• lychee_memory_consolidate

  Consolidate memory turns

  Try:Consolidate current session memoriesTry:Merge related memory entries

Comparable tools

Installation

pip install lycheemem

pip install "lycheemem[rerank]"

lycheemem-cli

{
  "mcpServers": {
    "lycheememory": {
      "command": "python",
      "args": ["-m", "lycheemem", "server"],
      "env": {}
    }
  }
}

Compare LycheeMem with