All tools
HarnessCommunity500

Claude Code Memory Bank

Updated Jul 7, 2026

A memory management system adapted from the Cline Memory Bank pattern, restructured for Claude Code by hudrazine. Uses six hierarchical markdown files — projectbrief, productContext, activeContext, systemPatterns, techContext, and progress — loaded via @import in CLAUDE.md. The /init-memory-bank command auto-detects technologies and creates only missing files, preserving any existing documentation.

View on GitHub

What it does

  • /init-memory-bank

    Smart initialization: detects existing project state, auto-infers tech stack, and creates only the missing memory bank files.

  • /workflow:understand

    Load all six memory bank files and summarize current project state at the start of a session.

  • /workflow:plan

    Draft a detailed implementation strategy grounded in the current activeContext.

  • /workflow:execute

    Implement the plan with systematic quality checks and update progress.md.

  • /workflow:update-memory

    Write the current session state back to all six hierarchical memory bank files.

Files (1)

OVERVIEW.mdprimary · markdown · 2.0 KB
## What it does

Brings the Cline Memory Bank pattern to Claude Code using six hierarchical markdown files stored in a memory-bank/ directory. The hierarchy flows from projectbrief.md through productContext.md and systemPatterns.md/techContext.md into activeContext.md, then into progress.md — each file builds on those above it so Claude always has a coherent picture of what the project is, how it is architected, and what is being worked on now. The core system is loaded into CLAUDE.md via @import so it is active in every session automatically.

The initialization command is smart: for a new project it prompts for details and creates all six files; for an existing project it reads package.json, README, and any existing documentation to auto-detect the technology stack and populate files without overwriting content that is already there; if a projectbrief.md already exists it reads it and uses it to populate the other five files consistently.

## Key commands

- /init-memory-bank — initialize the memory bank for the current project; detects existing state and fills only missing files
- /workflow:understand — load all six memory bank files and summarize the current project state and focus
- /workflow:plan — draft a detailed implementation strategy against the current activeContext
- /workflow:execute — implement the plan with systematic quality checks and update progress.md
- /workflow:update-memory — write the current session state back to all six memory bank files

## Install

    git clone https://github.com/hudrazine/claude-code-memory-bank.git
    cp -r claude-code-memory-bank/.claude /path/to/your/project/

Add to your project CLAUDE.md:

    @.claude/claude-memory-bank.md

Then in Claude Code run /init-memory-bank to populate the memory bank.

## When to use

Use when you want structured, hierarchical project memory that auto-detects the tech stack, preserves existing documentation, and gives Claude a consistent starting context across sessions — without the manual effort of maintaining a single flat CLAUDE.md.