openbrainstore.

Privacy Policy

OpenBrainStore is a memory service: you (or an AI client acting for you) deliberately store text so it can be recalled later. That makes the honest version of this policy short — the product is your data, scoped to your identity, exportable and deletable at any time. Here is exactly what we hold and what we do with it.

What we store

• Memories you write — the text bodies your AI clients send via remember, plus the structure around them: type, tags, entities, key-value pairs, and links between memories.
• Vector embeddings of memory bodies, computed on our server (no third-party AI service sees your text) to power semantic recall.
• Your GitHub identity — the numeric GitHub user id (your tenant key) and username, obtained when you sign in with GitHub. We request only the read:user scope; we never see your GitHub password and cannot touch your repositories.
• Subscription records — if you buy a paid plan: billing status, tier, renewal date, and Stripe's customer/subscription identifiers. Your card number never touches our servers; payment details go directly to Stripe.
• Operational records — beta signup and contact-form submissions you send us (username, email, what you wrote), OAuth tokens, and aggregate retrieval metrics (lane timings and result counts — not the content of your queries).
• Anonymous usage counters — page views on the site and app are counted with a daily-rotating hash; we never store your IP address with them, and the rotation makes tracking you across days impossible by construction. No cookies, no third-party analytics.

Where it lives

Primary storage is a server operated by us on Hetzner (US region) — canonical memory files plus a Postgres database, isolated per tenant with row-level security enforced in the database itself: your identity can only ever read rows keyed to it, fail-closed. Encrypted nightly backups go to Cloudflare R2 object storage. All traffic is TLS in transit.

How long we keep it

• Memories: until you delete them. Deleting (forget) removes a memory from search immediately; the underlying file is tombstoned for a 30-day recovery window, then purged.
• Backups: rolling 14 daily + 4 weekly snapshots. Deleted data ages out of backups within ~5 weeks at most.
• Subscription records: kept while your account exists, and as required for tax/accounting after that.

Who else is involved

We use four processors, each seeing only what their job requires: GitHub (sign-in identity), Stripe (payments — they hold your card, we hold their reference ids), Hetzner (the server), and Cloudflare (backup storage). We do not sell, rent, share, or advertise on your data. Your memory content is never used to train any model, ours or anyone else's.

Your data is portable — by design, not by policy

The export tool produces a complete, browsable bundle of your memories in the Open Knowledge Format (plain Markdown + YAML) at any time, no request or waiting period needed. The same server you're using is source-available — if you leave, your export runs on your own machine unchanged.

Deleting everything

Delete individual memories with forget. For full account deletion (all memories, identity records, and entitlements), use the contact form — we complete it within 30 days, after which only aging backup snapshots remain until they cycle out.

Changes

If this policy changes materially, the effective date above changes, and active subscribers get an email before the change takes effect.

Contact

Questions, deletion requests, or anything else: openbrainstore.com/#contact.