Vec Memory MCP Server
Provides graph-based semantic memory storage and retrieval using vector embeddings and SQLite, enabling relationship creation between memories and similarity-based search through natural language.
README
Vec Memory MCP Server
An MCP (Model Context Protocol) server that provides graph-based semantic memory using SQLite vec0 and Ollama embeddings.
Features
- Semantic Search: Vector-based similarity search using embeddings
- Graph Relationships: Create and traverse relationships between memories
- Flexible Transport: Support for both stdio and SSE (HTTP) transports
- Flexible Storage: SQLite with vec0 extension for efficient vector operations
- MCP Integration: Standard MCP server for easy integration
Prerequisites
- Node.js 18+
- Ollama: Install from ollama.ai or:
- macOS:
brew install ollama - Linux:
curl -fsSL https://ollama.ai/install.sh | sh - Windows: Download from ollama.ai/download
- macOS:
Installation
From npm (recommended)
Add to your MCP client configuration:
{
"mcpServers": {
"memory": {
"command": "npx",
"args": ["vec-memory-mcp"]
}
}
}
The server will be automatically downloaded and started when needed.
Note: The server will automatically start Ollama if it's not already running and download the required embedding model if needed.
From source
npm install
npm run build
Usage
Start the MCP server with stdio transport (default):
npm start
# or
npm run dev
Start with SSE transport for HTTP clients:
npm run build && node dist/index.js --sse
# or custom port
npm run build && node dist/index.js --sse --port 8080
Run npm run build && node dist/index.js --help for all options.
Environment Variables
MEMORY_DB_PATH: Path to SQLite database (default:./memory.db)OLLAMA_BASE_URL: Ollama API URL (default:http://localhost:11434)OLLAMA_MODEL: Embedding model to use (default:nomic-embed-text)
MCP Tools
Memory Operations
add_memory: Store content with semantic embeddingget_memory: Retrieve memory by IDupdate_memory: Update memory content or metadatadelete_memory: Remove a memorysearch_memories: Semantic search across memories
Relationship Operations
add_relationship: Create relationships between memoriesget_relationships: Query relationships with filteringupdate_relationship: Modify relationship strength or metadatadelete_relationship: Remove a relationshipget_connected_memories: Find memories connected through relationships
Architecture
src/ollama.ts: Ollama management and embedding generationsrc/database.ts: SQLite database schema and vec0 integrationsrc/memory.ts: Core memory operations and graph traversalsrc/server.ts: MCP server implementationsrc/index.ts: Entry point and configuration
Requirements
- Node.js 18+
- SQLite with vec0 extension (automatically checked)
- Ollama (must be installed separately)
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。