Simple Memory Extension MCP Server

An MCP server for extending the context window/memory of AI agents, supporting the storage, retrieval, and management of memory content, including semantic search functionality.

Knowledge management and memory Developer tools #Memory Extension #Semantic Search #Context Management #MCP Service Local .TypeScript

rating : 2.5 points

downloads : 7.2K

update time : 2025-04-28

Open Site

What is the Simple Memory Extension MCP Server?

This is a server based on the Model Context Protocol (MCP), specifically designed to help AI agents extend their memory capabilities. It allows AI to store important information, key decision points, or any content worth remembering, and quickly retrieve it when needed.

How to use this service?

Simply start the server, and then let the AI agent store or retrieve information through specific commands. The server provides two ways to manage memory: key - value storage and semantic search.

Use cases

It is particularly suitable for long - term conversations, large project development, creative workflows that require continuous context, or any scenario where AI needs to remember a large amount of information.

Main Features

Context Item Management

Provides basic operations for storing, retrieving, and deleting key - value pairs.

Namespace Management

Allows creating different namespaces to organize memory content.

Semantic Search

Semantic search function based on the E5 model, which can find content by meaning rather than exact match.

Advantages

Extend the memory capabilities of AI agents, breaking through the context window limit.

Support semantic search, so relevant content can be found even if the exact keywords are not remembered.

Lightweight and easy to deploy.

Limitations

The semantic search function requires Python dependencies.

Currently only supports SQLite as the backend storage.

In the early stage of development, the functions may not be fully developed.

How to Use

Install Dependencies

Ensure that the Node.js environment is installed, and then install the project dependencies.

Start the Server

Use the following command to start the MCP server.

Configure Environment Variables

Edit the configuration in the.env file as needed.

Usage Examples

Long - term Project Development

When developing a large project, AI can store key design decisions and technology choices, and refer to them at any time later.

Creative Writing Assistance

In long - form creative writing, AI can remember character settings and plot key points to maintain story consistency.

Frequently Asked Questions

How does the semantic search function work?

Where is the data stored?

How to expand the storage capacity?

Related Resources

Model Context Protocol Documentation

Official documentation about the MCP protocol.

E5 Model Introduction

The semantic search embedding model used.

GitHub Repository

Project source code.

🚀 Simple Memory Extension MCP Server

An MCP server designed to extend the agent's context window/memory. It's incredibly useful when coding large - scale features or engaging in vibe coding, allowing you to store/retrieve progress, critical moments, changes, or anything else worth remembering. Simply instruct the agent to store memories and retrieve them when needed, or let the agent manage its own memories (through cursor rules, etc.).

🚀 Quick Start

Start the Server

npm install
npm start

Available Tools

Context Item Management

store_context_item - Store key - value pairs in a namespace.
retrieve_context_item_by_key - Get the value by key.
delete_context_item - Delete a key - value pair.

Namespace Management

create_namespace - Create a new namespace.
delete_namespace - Delete a namespace and all its contents.
list_namespaces - List all namespaces.
list_context_item_keys - List keys in a namespace.

Semantic Search

retrieve_context_items_by_semantic_search - Find items based on meaning.

💻 Usage Examples

Basic Usage

# Start the server
npm install
npm start

Advanced Usage

# Use semantic search to find relevant context items
# This requires the semantic search feature to be set up correctly

🔧 Technical Details

Semantic Search Implementation

Use the E5 model to convert queries into vectors.
Automatically split text into chunks for better matching.
Calculate the cosine similarity between the query and stored chunks.
Filter results based on a threshold and sort them by similarity.
Return the top - matching items and their full item values.

📦 Installation

Development

# Development server
npm run dev

# Format code
npm run format

.env Configuration

# Database file path
DB_PATH=./data/context.db

PORT=3000

# Use HTTP SSE or standard input/output
USE_HTTP_SSE=true

# Log level: debug, info, warn, error
LOG_LEVEL=info

📚 Documentation

Semantic Search Explanation

This project includes a semantic search feature implemented using Hugging Face's E5 embedding model. This enables you to find content based on the meaning of context items rather than exact key matching.