Structured project knowledge management for LLMs via Model Context Protocol (MCP)
---
> **Note:** This is not a traditional Node.js application. Memory Bank MCP is an **MCP server**—a component in the [Model Context Protocol](https://modelcontextprotocol.io/introduction) ecosystem. It exposes project knowledge to LLM-powered agents and tools using a standardized protocol, enabling seamless integration with AI clients (e.g., Claude Desktop, IDEs, or custom LLM agents).
---
## What is Model Context Protocol (MCP)?
MCP is an open protocol that standardizes how applications provide context to LLMs. Think of MCP like a USB-C port for AI: it provides a universal way to connect AI models to data sources and tools, both locally and remotely. MCP enables:
- **Plug-and-play integrations** between LLMs, data, and tools
- **Switching between LLM providers** with minimal friction
- **Secure, modular architecture** for building AI workflows
Learn more: [MCP Introduction](https://modelcontextprotocol.io/introduction)
## About Memory Bank MCP
Memory Bank MCP is an **MCP server** that helps teams create, manage, and access structured project documentation. It generates and maintains interconnected Markdown documents capturing all aspects of project knowledge, from high-level goals to technical details and daily progress. It is designed to be accessed by MCP-compatible clients and LLM agents.
## Features
- **AI-Generated Documentation**: Uses Gemini API to generate and update project documentation
- **Structured Knowledge System**: Maintains six core document types in a hierarchical structure
- **MCP Server**: Implements the Model Context Protocol for integration with LLM agents and tools
- **Customizable Storage**: Choose where your Memory Bank directory is created
- **Document Templates**: Pre-defined templates for project brief, product context, system patterns, etc.
- **AI-Assisted Updates**: Update documents manually or regenerate them with AI
- **Advanced Querying**: Search across all documents with context-aware relevance ranking
## Installation
```bash
# Clone the repository
git clone https://github.com/tuncer-byte/memory-bank-mcp.git
cd memory-bank-mcp
# Install dependencies
npm install
# (Optional) Create .env file with your Gemini API key
echo "GEMINI_API_KEY=your_api_key_here" > .env
```
## Usage
> **Note:** Memory Bank MCP is intended to be run as an MCP server, not as a standalone app. You typically launch it as part of an MCP workflow, and connect to it from an MCP-compatible client (such as Claude Desktop or your own LLM agent).
### Development Mode
```bash
npm run dev
```
### Production Mode
```bash
npm run build
npm run start
```
### MCP Integration
To connect Memory Bank MCP to your MCP client, add the following to your `mcp.json` configuration:
```json
{
"memoryBank": {
"command": "node",
"args": ["/path/to/memory-bank-mcp/dist/index.js"],
"env": {
"GEMINI_API_KEY": "your_gemini_api_key_here"
}
}
}
```
Replace `/path/to/memory-bank-mcp/dist/index.js` with the absolute path to your built file, and add your Gemini API key if needed.
---
## MCP Tools Exposed by Memory Bank
Memory Bank MCP provides the following tools via the Model Context Protocol:
### `initialize_memory_bank`
Creates a new Memory Bank structure with all document templates.
**Parameters:**
- `goal` (string): Project goal description (min 10 characters)
- `geminiApiKey` (string, optional): Gemini API key for document generation
- `location` (string, optional): Absolute path where memory-bank folder will be created
**Example:**
```javascript
await callTool({
name: "initialize_memory_bank",
arguments: {
goal: "Building a self-documenting AI-powered software development assistant",
location: "/Users/username/Documents/projects/ai-assistant"
}
});
```
### `update_document`
Updates a specific document in the Memory Bank.
**Parameters:**
- `documentType` (enum): One of: `projectbrief`, `productContext`, `systemPatterns`, `techContext`, `activeContext`, `progress`
- `content` (string, optional): New content for the document
- `regenerate` (boolean, default: false): Whether to regenerate the document using AI
**Example:**
```javascript
await callTool({
name: "update_document",
arguments: {
documentType: "projectbrief",
content: "# Project Brief\n\n## Purpose\nTo develop an advanced and user-friendly AI..."
}
});
```
### `query_memory_bank`
Searches across all documents with context-aware relevance ranking.
**Parameters:**
- `query` (string): Search query (min 5 characters)
**Example:**
```javascript
await callTool({
name: "query_memory_bank",
arguments: {
query: "system architecture components"
}
});
```
### `export_memory_bank`
Exports all Memory Bank documents.
**Parameters:**
- `format` (enum, default: "folder"): Export format, either "json" or "folder"
- `outputPath` (string, optional): Custom output path for the export
**Example:**
```javascript
await callTool({
name: "export_memory_bank",
arguments: {
format: "json",
outputPath: "/Users/username/Documents/exports"
}
});
```
## Document Types
Memory Bank organizes project knowledge into six core document types:
1. **Project Brief** (`projectbrief.md`): Core document defining project objectives, scope, and vision
2. **Product Context** (`productContext.md`): Documents product functionality from a user perspective
3. **System Patterns** (`systemPatterns.md`): Establishes system architecture and component relationships
4. **Tech Context** (`techContext.md`): Specifies technology stack and implementation details
5. **Active Context** (`activeContext.md`): Tracks current tasks, open issues, and development focus
6. **Progress** (`progress.md`): Documents completed work, milestones, and project history
## License
MIT
