home / mcp / memvid mcp server

Memvid MCP Server

Provides a persistent, file-based memory store for Memvid that supports project-based memory, content addition, semantic search, and optional natural language querying.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "khgs2411-memvid_mcp": {
      "command": "npx",
      "args": [
        "-y",
        "memvid-mcp-server@latest"
      ],
      "env": {
        "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY",
        "MEMVID_LOCAL_STORAGE": "0"
      }
    }
  }
}

You can run a Memvid MCP Server to give AI agents persistent, file-based memory. It creates isolated memory projects, lets you store content and metadata, and enables fast semantic search across your memories. Optional natural language querying lets you ask questions about your data when you provide an API key.

How to use

Use this MCP server with any MCP-compatible client to manage memory per project. Start a memory project with create_or_open_memory, add content with add_content, and search memory with search_memory to quickly find relevant items. If you provide an API key, you can also use ask_memory to query your data in natural language.

Key capabilities you can leverage include: storing text documents, code snippets, and metadata; organizing memories per project using .mv2 files; performing semantic search through a hybrid of lexical and semantic techniques; and optionally asking natural language questions about stored content.

How to install

Prerequisites: you need a runtime capable of running Node.js tools or Bun for package installation.

Install the MCP server globally using npm or Bun.

npm install -g memvid-mcp-server
# or
bun add -g memvid-mcp-server

Configuration and usage notes

You can configure the server to work with any MCP-compatible client. The following environment variables control optional features and storage location:

OPENAI_API_KEY enables the optional ask_memory tool and may influence embedding behavior in some modes. MEMVID_LOCAL_STORAGE toggles local storage location.

Example client configuration (Claude Desktop) to run the MCP server via npx with environment variables set:

{
  "mcpServers": {
    "memvid": {
      "command": "npx",
      "args": ["-y", "memvid-mcp-server@latest"],
      "env": {
        "OPENAI_API_KEY": "sk-...",
        "MEMVID_LOCAL_STORAGE": "0"
      }
    }
  }
}

Available tools

create_or_open_memory

Initialize a new memory project or open an existing one to begin adding content.

add_content

Store text, code, and metadata into the memory project for later retrieval.

search_memory

Query stored content with lexical or semantic search to find relevant items.

ask_memory

Optional tool to ask natural language questions about your memory using an LLM.