memory-bank-mcp

Memory Bank MCP Server

Project context, progress notes, and decision records live in Markdown files with memory-bank-mcp. It gives AI assistants tools to maintain that information across sessions, serving developers working with local projects or shared storage over SSH.

Server facts

memory-bank-mcp

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
35
Last commit
Data as of

Tags

  • Memory Bank
  • mcp
  • remote ssh

Overview

Memory-bank-mcp organizes project knowledge into Markdown records covering goals, current work, progress, decisions, and architectural patterns. An assistant can initialize a repository, select its location, inspect available files, and read or revise their contents. Dedicated tools record progress, refresh active context, and capture decisions, including their background and alternatives. Status checks and configuration diagnostics help inspect the setup, while a migration tool converts file names from camelCase to kebab-case. These operations support development workflows where project information needs to remain available beyond a single assistant session.

The mode system distinguishes coding, architecture, questions, debugging, and testing work. Assistants can change modes and inspect the current mode; project .clinerules files also support mode-specific behavior. Separate tools process and complete Update Memory Bank requests for temporary updates during a session. Storage can stay in a configurable project folder or reside on a remote machine reached through SSH. Remote use requires key authentication and permission to modify the destination directory; password authentication is unsupported. The Node-based server communicates over stdio, and the documentation describes integration with Cursor.

memory-bank-mcp setup notes

  • memory-bank-mcp records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers