Memory Bank

Memory Bank MCP lets you manage Memory Banks for AI assistants, enabling you to store, retrieve, and track context across sessions. It provides a structured set of tools to initialize banks, read and write files, log progress and decisions, and switch between focused modes like code or architect tasks. With Memory Bank MCP you gain persistent context, better collaboration between sessions, and clear visibility into the current state of your memory repository.

How to use

You interact with Memory Bank MCP through a client that talks MCP to the server. Start by initializing or loading a Memory Bank, then perform actions to update context, track progress, and log decisions as your project evolves. Use mode switching to tailor the assistant’s focus: code, architect, ask, debug, or test. When you want quick updates or temporary changes during a session, you can apply the UMB command to update files temporarily.

How to install

Prerequisites: you need Node.js and npm or npx available on your system.

# Install Memory Bank MCP locally
npm install @movibe/memory-bank-mcp

# Or install globally
npm install -g @movibe/memory-bank-mcp

# Or run directly with npx (no installation required)
npx @movibe/memory-bank-mcp
```,

Additional content and usage notes

Configuration, usage patterns, and practical examples are provided through CLI commands and arguments. Common actions include initializing a Memory Bank, tracking progress, logging decisions, and switching modes. You can operate Memory Bank MCP via npx to run a specific mode or invoke it directly through its CLI entry points. The Memory Bank structure includes sections like product-context.md, active-context.md, progress.md, decision-log.md, and system-patterns.md, which organize information consistently across sessions.

MCP Server connection details

You can connect to the Memory Bank MCP via a direct local process (stdio) or by invoking it through a remote-like CLI flow. The standard approach shown here uses the local CLI to run commands that interact with a Memory Bank in your project.

Available tools

initialize_memory_bank

Create and prepare a Memory Bank in your project, setting up the memory-bank directory and baseline files for product context, active context, progress, decision log, and system patterns.

track_progress

Record progress updates with an action and optional description to document milestones and ongoing work within the Memory Bank.

log_decision

Capture important decisions with a title, contextual information, and the rationale behind the choice, including alternatives considered.

switch_mode

Change the active operating mode to focus on code, architect, ask, debug, or test tasks to tailor the assistant’s behavior.

Umb_command

Temporarily update Memory Bank files during a session without committing changes, enabling experimentation and exploration.