This project provides a Model Context Protocol (MCP) server for centralized, remote management of AI memory banks, inspired by Cline Memory Bank. It targets developers and AI users seeking to manage memory banks across multiple projects, offering features like project isolation, path traversal prevention, and a full MCP protocol implementation for remote file operations.

How It Works

The server acts as a central hub, transforming local file-based memory banks into a network service. It enforces a consistent file structure and provides project-specific directories, ensuring isolation and preventing path traversal. The implementation supports core MCP operations such as reading, writing, updating, and listing memory bank files and projects, with configurable auto-approval for specific operations.

Quick Start & Requirements

• Install: npx -y @smithery/cli install @alioshr/memory-bank-mcp --client claude
• Prerequisites: Node.js, npm. Configuration requires updating specific cline_mcp_settings.json files depending on the client (Cursor, Roo Code, Claude Desktop).
• Configuration: Requires setting MEMORY_BANK_ROOT to the desired storage directory.
• Docs: Custom IA instructions

Highlighted Details

• Full MCP protocol implementation for remote memory bank management.
• Multi-project support with enforced file structure and path traversal prevention.
• Optional Docker image for containerized deployment.
• Type-safe operations and proper error handling.

Maintenance & Community

• Development commands include npm install, npm run build, npm run test, and npm run dev.
• Contributions are encouraged via standard Git workflow (fork, branch, commit, PR).
• Development guidelines emphasize TypeScript, type safety, testing, and documentation.

Licensing & Compatibility

• MIT License. Permissive for commercial use and closed-source linking.

Limitations & Caveats

The project is inspired by Cline Memory Bank, suggesting potential reliance on or compatibility with that ecosystem. Specific configuration paths are provided for certain clients, implying potential setup variations for other environments.