Claude Code · Cursor · Codex — catches drift before you re-litigate
When your AI undoes a call you already made, it says Hold on — and why.
"Hold on — you dropped Lemon Squeezy last week: it won't pay out where you live. Shipping it Friday breaks checkout. Want Stripe instead?"
Same memory across tools on your machine — so Cursor, Claude Code, and Codex don't re-open decisions you already closed. Local-first. Free on your machine; Pro adds sync and AI included.
npx -y window-memory init
Don't want to touch a terminal? Paste it into Cursor or Claude Code and let it set itself up.
Works with Claude Code, Cursor, and Codex. One line to set up. Free.
Local-first. Free on your machine — your key, your data. Pro adds sync and AI included.
Your AI reaches for something you already threw away, or drifts off your goal. Window Memory catches it — before it ships.
Guarding memory: past decisions and failures stay in play, so your AI doesn't ship what you already threw away. Open the dashboard any time to see each decision and why.
Tomorrow, just say hi. Your AI is already caught up — what you changed, why you changed it, and what's left before your deadline.
Your repo shows what you built — not why you rejected what you already tried. Open a different tool and Window Memory still knows why you made each call — the part that's in no file — so your AI never re-litigates a decision you already made.
Reading your files tells your AI what the code is today — not why you chose this over the thing you already tried and dropped. That reasoning lives only in Window Memory, and it follows you into every tool.
Nothing to set up. Nothing to maintain.
Your project's goal is fed to your AI on its own — every session, and again every time it gets back to work. Set it once or let Window Memory draft it from your work, then steer it from the dashboard. You never sit down to configure anything. Install once and it starts on its own after every reboot — running quietly in the background, nothing to restart or babysit.
I'm not a developer by background. I build with AI tools — Claude Code, Cursor, Codex — and the thing that hurt most was this: my AI forgot everything between sessions.
I'd go in circles without realizing it — retrying things that had already failed, undoing decisions I'd made for a reason, burning tokens. By the time I noticed I was looping, I didn't even know where to start fixing it. If you didn't write the code yourself, that's a rough place to be.
So I built Window Memory. It doesn't just look at git or your code — it remembers how your project was actually built: the intent, the reason, the result, and the failures behind each step. So your AI keeps the context, and stops walking you back into the same wall.
One line and the loop-breaker is on. Your AI remembers the why behind every decision — your key, your data, your computer.
Pick up on any device, watch new decisions land live on your phone, and — across the projects you keep on Pro — ask your AI to read a past project and apply it here.
On Pro, invite teammates — everyone's builds pool in one place. Who did what, which tool, every wall they hit. Your AI avoids the whole team's failures, not just yours.
Run npx -y window-memory init — one line, once.
Keep working in Claude Code, Cursor, or Codex. Nothing changes.
Open your next session — your AI already knows where you stopped, and why.
It keeps running on its own after every reboot. Install once, never think about it again.
Not a walled garden. See what your AI decided, one card at a time — no code to read. Pro syncs across devices; power users can plug it into their own tools.
Every decision as a card — intent, reason, result. Follow what your AI actually did.
Claude Code, Cursor, and Codex pull context in automatically — search by meaning, not keywords.
Issue safe keys for Zapier, n8n, or your own scripts — if you want them.
Take over a teammate's work and your AI already knows the whole story — every decision they made, and every wall they already hit. Handoffs in minutes, not weeks.
Free is one active project on your machine. Pro follows you to every device — open it on your phone, anywhere.
Free forever
One active project on Free — perfect to try on one machine.
$15/seat/mo
or $120/yr — early-bird, first 100 (then $144/yr)
Try Pro free for 3 days — no key, no card.
Does it see my code?
No. Sensitive data is stripped before anything leaves your computer. On Free it runs on your own AI key; on Pro the analysis is included — no key to manage — and we never keep your raw sessions.
Do I have to keep it running, or restart it?
No. It starts on its own every time you turn your computer on and runs quietly in the background. You install it once.
I installed it but my first session didn't catch up?
If your editor (Claude Code, Cursor, Codex) was already open when you installed, just restart it once — it picks up Window Memory on the next session. After that, every new session is caught up automatically.
Will it touch my CLAUDE.md or existing setup?
It never edits your project's CLAUDE.md or repo rules. Setup adds a small Window Memory channel (removable with npx window-memory uninstall). Your project files stay yours.
Can it use what I figured out in my other projects?
Yes — on Pro, where you keep all your projects (Free keeps one). Each project's memory stays separate by default; nothing auto-mixes. But you can pull across the projects you've built with Window Memory on demand: search a past one and bring a decision or lesson into the one you're building now. You stop re-solving the same problem in every new repo.
What does it cost?
Free on your own computer — bring your own AI key. Pro ($15/mo) includes the AI — no key to manage — and follows you across every device and your team. New accounts start with a 3-day Pro trial — no key, no card.
Can my own tools and automations read it?
Yes. Your memory isn't a walled garden — connect it to the tools you already use. Your AI tools read it directly, and you can issue safe, one-click keys for Zapier, n8n, or any automation you set up.
Can I cancel or get a refund?
Pro subscribers: open the dashboard → Settings → Plan & billing → Manage billing. That opens Polar, where you can cancel, update payment, or download receipts. Cancel anytime — access continues until the end of your paid period, then you're back on Free with your local data. Refunds follow Polar's policy; if something went wrong, email hello@windowmemory.com.
Need help?
Email hello@windowmemory.com — a real person (the developer) reads every message.
npx -y window-memory init
Free on your computer (your own key) · Pro $15/mo, AI included · 3-day Pro trial.
Built with itself — every feature shipped while Window Memory remembered its own development.