MCP 服务器

haksnbot-memory

Provides persistent memory storage for AI agents with full-text search, tagging, and importance levels, enabling agents to store and retrieve memories efficiently.

README

haksnbot-memory

An MCP (Model Context Protocol) server that provides persistent memory storage for AI agents. Uses SQLite with full-text search for efficient storage and retrieval of memories.

Part of the Haksnbot suite: This project was originally developed as part of Haksnbot, an autonomous Minecraft bot. The suite includes four repos that work together: haksnbot-tools (Minecraft bot control), haksnbot-mind (the autonomous agent), haksnbot-admin (server administration), and haksnbot-memory (this repo - persistent memory). Each can be used independently, but they're designed to work together.

Features

Persistent storage - Memories survive agent restarts
Full-text search - Find memories by content using FTS5
Tagging system - Organize memories with flexible tags
Importance levels - Prioritize significant memories
Pagination - Handle large memory collections efficiently

Installation

git clone https://github.com/haksndot/haksnbot-memory.git
cd haksnbot-memory
npm install

Usage with Claude Code

Add to your Claude Code MCP settings (~/.claude/settings.json):

{
  "mcpServers": {
    "memory": {
      "command": "node",
      "args": ["/path/to/haksnbot-memory/src/index.js"]
    }
  }
}

Configuration

The database path can be set via environment variable:

{
  "mcpServers": {
    "memory": {
      "command": "node",
      "args": ["/path/to/haksnbot-memory/src/index.js"],
      "env": {
        "MEMORY_DB_PATH": "/path/to/memories.db"
      }
    }
  }
}

If not set, defaults to memories.db in the package directory.

Available Tools

Tool	Description
`create_memory`	Store a new memory with optional tags, context, and importance
`get_memory`	Retrieve a specific memory by ID
`update_memory`	Update an existing memory's content, tags, or importance
`delete_memory`	Remove a memory
`list_memories`	List memories with filtering, sorting, and pagination
`search_memories`	Full-text search across memory content, tags, and context
`get_memory_stats`	Get statistics: total count, tag distribution, date range

Memory Structure

Each memory contains:

content - The main memory text (required)
tags - Array of categorization tags (e.g., ["player:Steve", "event:trade"])
context - Additional context about when/where the memory was created
importance - Priority level 1-10 (default 5)
created_at / updated_at - Timestamps

Examples

> Remember that Steve helped me find diamonds at coordinates -234, 12, 567

> Search my memories for anything about Steve

> List my most important memories (importance >= 8)

> What do I remember about trading?