Libram

Privacy policy

Effective September 2, 2026

This policy explains what information Libram collects, how it is used, and the choices you have. It is written to be read, not skimmed past — it's short because we collect little.

Who this covers

Libram is self-hostable software. This policy applies to the services we operate: this website (libram.dev), the public demo workspace (demo.libram.dev), the sign-in relay (auth.libram.dev), and any other Libram workspace we run ourselves.

If you use a Libram workspace that someone else hosts on their own Cloudflare account, that operator controls the data in it and is responsible for their own privacy practices. Our sign-in relay may still be involved in getting you signed in to such a workspace (see below), and this policy covers that step.

Information we collect

When you sign in with Google or GitHub

Libram has no passwords; you sign in with an existing account. We request only basic profile information:

ProviderWhat we ask forWhat we receive
GoogleThe openid, email, and profile scopesYour Google account ID, email address (if verified by Google), display name, and profile picture URL
GitHubThe read:user and user:email scopesYour GitHub user ID, username, display name, avatar URL, and verified email addresses

We do not request, and cannot access, anything else in your Google or GitHub account — no files, mail, calendar, contacts, repositories, or activity. We do not store the access or refresh tokens these providers issue; they are used once, during sign-in, to fetch the profile above, and then discarded.

We use this information to:

The sign-in relay (auth.libram.dev)

Instances we operate share a single Google and GitHub app registration. To make that work, the sign-in round trip runs on auth.libram.dev, which then passes a short-lived, signed identity (the profile fields above) to the workspace you're signing into. The relay has no database: it stores nothing about you beyond the cookies needed to complete a single sign-in, which expire within minutes. It only delivers identities to origins on an explicit allowlist.

Content you create

Documents, comments, suggestions, uploaded files, folder structure, and the activity log that records who changed what. This is the product; storing it is the point. It is stored in the Cloudflare account that hosts the workspace, and is visible to the other members of that workspace.

Docs you deliberately publish are visible to anyone with the link, and may be indexed by search engines. Files you upload are served at unguessable URLs so that they can be embedded in published docs; anyone who has such a URL can open the file.

Automatic summaries

Libram generates short summaries and keywords for docs and folders to make search and navigation better. To do this, document content is processed by a language model running on Cloudflare Workers AI, within the same Cloudflare account as the workspace. Summaries are generated automatically and are not editable. This feature can be turned off by the workspace operator.

Agents and integrations you connect

Libram exposes a REST API and an MCP (Model Context Protocol) server so that AI tools can work with your docs. Those tools act only with credentials you create — a personal API token, or an OAuth authorization you grant in the browser with a scope you choose (read, comment, or write). What a connected tool then does with the content it reads is governed by that tool's own privacy policy. You can revoke a token or authorization at any time from the workspace settings, and every action taken through one is attributed in the activity log as automated, so it's always clear what a person did and what a tool did.

Technical information

Like any web service, our servers see your IP address, browser type, and the pages you request. Cloudflare, our hosting provider, processes this to serve the site and defend against abuse. We do not run third-party analytics or advertising trackers on these sites.

Cookies

We use a session cookie to keep you signed in to a workspace, and short-lived cookies to protect the sign-in flow against forgery. No cookies are used for tracking or advertising.

How we share information

We do not sell your information, and we do not share it with third parties for their own marketing. Your information is disclosed only:

Google API Services — Limited Use

Libram's use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. Concretely: we use the Google profile information above only to provide and improve the sign-in and collaboration features of Libram that you see; we don't use it for advertising, we don't sell it, and humans don't read it except with your consent, for security purposes, to comply with the law, or in aggregated, anonymized form.

Retention and deletion

Security

All traffic to our services is encrypted in transit. Data is stored on Cloudflare's infrastructure, which is encrypted at rest. Session cookies are HTTP-only and secure. API tokens are stored only as hashes. Sign-in identities passed between the relay and a workspace are signed and single-use.

Children

Our services are not directed to children under 13 (or the age of digital consent where you live), and we don't knowingly collect information from them.

International transfers

Cloudflare operates a global network, so your information may be processed in countries other than your own. Cloudflare maintains appropriate safeguards for these transfers.

Changes

If we change this policy in a way that matters, we'll update the date at the top and, for changes that affect how we use your information, notify workspace members. Continued use after a change means you accept it.

Contact

Questions or requests about your information: privacy@libram.dev.