home / mcp / claude conversation memory mcp server

Claude Conversation Memory MCP Server

Provides a local MCP server that indexes, searches, and summarizes Claude conversation history for context-aware retrieval.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "adamkwhite-claude-memory-mcp": {
      "command": "sh",
      "args": [
        "-c",
        "cd $HOME/Code/claude-memory-mcp && python3 src/server_fastmcp.py"
      ]
    }
  }
}

You can store Claude conversation history locally and search it, extract topics automatically, and generate weekly summaries to gain quick insights. This MCP server provides fast, persistent context for Claude Desktop, so your conversations stay organized and searchable across sessions.

How to use

You connect this MCP server to Claude Desktop to enable persistent memory for your conversations. Once linked, you can search your stored conversations by topic or content, rely on automatic topic categorization to organize entries, and generate weekly summaries that reveal patterns and insights. Use the server from your Claude Desktop MCP configuration to ensure your memory is available during current and future sessions.

How to install

Prerequisites you need before installation are Python 3.11 or newer and a suitable environment for running Claude Desktop with MCP integration.

# Option 1: Quick stdio install via Claude Code (recommended)
claude mcp add --transport stdio claude_memory -- sh -c "cd $HOME/Code/claude-memory-mcp && python3 src/server_fastmcp.py" 

# Replace the path with the actual location where you cloned the project

# Example variants for different clone locations

# If cloned to ~/Code/claude-memory-mcp (default)
claude mcp add --transport stdio claude_memory -- sh -c "cd $HOME/Code/claude-memory-mcp && python3 src/server_fastmcp.py"

# If cloned to ~/projects/claude-memory-mcp
claude mcp add --transport stdio claude_memory -- sh -c "cd $HOME/projects/claude-memory-mcp && python3 src/server_fastmcp.py"

# If cloned to ~/dev/claude-memory-mcp
claude mcp add --transport stdio claude_memory -- sh -c "cd $HOME/dev/claude-memory-mcp && python3 src/server_fastmcp.py"

Additional sections

Configuration notes, usage tips, and troubleshooting help are provided below to ensure you can get the MCP server running and keep it stable.

Configuration

Storage, logging, and environment settings influence how the memory store behaves. You can customize storage location and logging format to suit your environment.

Troubleshooting

If you encounter issues, verify that the MCP server is reachable by your Claude Desktop client and that the local memory path is writable. Check that the Python environment uses Python 3.11+ and that required dependencies are installed (see installation steps). If you see search results that don’t appear, validate the index and topic extraction setup and ensure your conversations were properly imported.

Notes

The MCP server is designed to integrate with Claude Desktop to provide persistent conversation context. It supports full-text search, topic extraction, weekly summaries, and organized storage by date and topic, with fast retrieval via a relevance-based approach.

Available tools

search_conversations

Search stored conversations by topic or content with optional result limits to quickly locate relevant context.

add_conversation

Add a new conversation entry to memory with content, title, and date for persistent storage.

generate_weekly_summary

Create a summary report of conversations for the current or a past week, highlighting insights and patterns.