Memory Engineering

Provides persistent memory and semantic code understanding for AI-assisted code work through MongoDB and Voyage AI embeddings.

javascript

11

GitHub Stars

javascript

Language

6 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "romiluz13-memory-engineering-mcp": {
      "command": "npx",
      "args": [
        "memory-engineering-mcp"
      ],
      "env": {
        "MONGODB_URI": "YOUR_MONGODB_ATLAS_URI",
        "VOYAGE_API_KEY": "YOUR_VOYAGE_API_KEY"
      }
    }
  }
}

You run Memory Engineering MCP to enable semantic code memory and search capabilities for your AI assistants. It uses MongoDB Atlas for vector storage and Voyage AI embeddings to understand code patterns, making it easy to locate related code, track decisions, and reuse context across your project.

How to use

You interact with Memory Engineering MCP through a lightweight client that starts a local MCP server or connects to a remote MCP endpoint. Start the MCP runner, ensure your environment variables are set, and then use the MCP commands to initialize memories, search code semantically, and update memory entries as you work.

How to install

Prerequisites: you need Node.js and npm installed on your machine. You will also need access to a MongoDB Atlas instance and a Voyage API key.

Additional setup notes

// Minimal local startup configuration for Memory Engineering MCP as a stdio server
{
  "mcpServers": {
    "memory_engineering": {
      "command": "npx",
      "args": ["memory-engineering-mcp"],
      "env": {
        "MONGODB_URI": "your-mongodb-atlas-uri",
        "VOYAGE_API_KEY": "your-voyage-api-key"
      }
    }
  }
}

First run and common workflows

Install the MCP global package, then run the MCP runner to start a local server. After that you can initialize memories, perform semantic searches, and update memory entries as you work.

Key commands you will use:

Initialize: memory_engineering_init
Semantic search: memory_engineering_search --query "your query" --codeSearch "pattern"
Update a memory: memory_engineering_memory --name activeContext --content "Your updated context here"

Security and credentials

Provide secure access to your MongoDB Atlas cluster and manage Voyage API credentials. Do not expose MONGODB_URI or VOYAGE_API_KEY in public or shared environments. Rotate credentials regularly and use environment-specific configurations.

Troubleshooting

If memory embeddings fail to load or searches return no results, check that the MongoDB connection is reachable, the Atlas embeddings are enabled, and your Voyage API key is valid. Verify that environment variables are correctly passed to the MCP process and that the MCP command is running in the expected working directory.

Notes

This MCP relies on 1024-dimension embeddings stored in MongoDB Atlas and uses 32K token context windows for long-file understanding. Ensure your MongoDB cluster has sufficient capacity for the size of your codebase and that your network allows vector search operations.

Available tools

memory_engineering_init

Initializes the project by scanning code, creating memories, and generating embeddings.

memory_engineering_search

Performs semantic search across memories and code using natural language and code patterns.

memory_engineering_memory

Reads or updates memories across the seven core memory categories.

memory_engineering_sync

Syncs code embeddings incrementally to reflect changes in the codebase.

memory_engineering_system

Runs health checks and diagnostics for the MCP environment.