Memory for Claude Code is a companion app for your Mac. It stores notes, decisions, and project information in a local database that Claude reads at the start of each session.
Work on a project for months — code, writing, research, design, anything. Take a break. Come back weeks later. Claude remembers exactly where you left off.
Use it your way. Track bugs and features, or journal your creative process. The memory is yours to shape.
We both write. We both read.
With the optional session hook from the Help Book, a briefing like this lands in Claude's context before your first prompt. You never see it on screen — Claude does.
$ claude ## Project memory briefing (auto-generated at session start) ### Essentials (starred) - #12 Auth: JWT with 15-minute tokens + refresh (decided) - #31 Ship checklist: metadata done, screenshots pending ### Open deadlines - #27 Onboarding flow [due: 2026-08-15] ### Latest records - #34 2026-07-28 — Search pagination bug fixed - #33 2026-07-26 — Tag filters: multi-select shipped - #32 2026-07-24 — Refactor: exports module simplified
Take a break. Come back weeks later. Ask "where were we?" — and watch Claude answer from memory.
▸ In the Finder, create a folder: "Master Thesis" ▸ Open the folder in Memory for Claude Code pick "Academic" — the memory is born $ cd ~/Desktop/Master\ Thesis $ claude ## Project memory briefing #1 Welcome — how this project remembers › Read the CLAUDE.md ▸ Claude learns the rules, starts writing records › Let's start by updating record #1 ▸ #1 updated — the project has its context $ claude (weeks later) › where were we? ▸ #23 Chapter 2 approved · next: literature review ▋
This is not a log you are handed — it is a shared workspace, with rights on both sides.
Claude records as it works; you open the app and edit — correct a record, star what matters, archive what is done. Records you touch become protected: Claude asks before ever changing your words. If you both edit the same record, a banner lets you choose which version survives.
Every project has a mailbox. The Claude working on your app can leave a letter for the Claude working on your website — waiting at the start of that project's next session. When a new rule applies everywhere, one of them spreads the word.
Project deadlines and per-task due dates, with a chart that shows the remaining days at a glance. Claude works at super speed — the deadlines keep the human side of the team on track.
Memory for Claude Code creates a SQLite database in your project's .claude folder. When you start a Claude Code session, it reads the relevant memories to maintain context.
Yes. All data is stored locally on your Mac. No cloud, no accounts, no data collection.
Absolutely. Writing, research, design, architecture — anything you work on with Claude Code. The app ships with 13 project types — from code to academic, journalism to travel — each with a ready-made set of 12 tags, and every tag can be renamed to fit your workflow.
The memory database is a standard SQLite file. Back it up with your project or sync with Git.
The memory lives inside your project folder, so it travels with the project — sync the .claude folder with Git and every machine (and every Claude) reads the same memory.
Memory for Claude Code is a one-time purchase — yours forever. However, it works with Claude Code, which requires a Claude Pro subscription or higher.
