We built a bookmark manager we don’t need to be trusted with.

What you save is a fairly complete picture of what you are reading, buying, worrying about, and working on. Most bookmark managers hold that picture in readable form on their servers, and most of them mean well. But meaning well is a policy, and policies change with funding rounds, acquisitions, subpoenas, and breaches.

So Bracemark is built the other way round. Your password unlocks a key on your own device; every link, list, and tag becomes its own encrypted file before it is uploaded; what the server keeps of it is a path, a size, a timestamp, and some bytes. Around that sits the account — your username, your sessions, your plan if you buy one — and nothing readable from a single link. There is no “we promise not to look” in that sentence, because there is no looking available to promise against.

Bracemark is the successor to Brace.to, a bookmark manager we have run since 2020 and still run today, on a decentralised identity stack. That one worked, and the people who used it were generous with their patience — but it was named after a domain nobody could say out loud, and better ways to do the same job arrived. Nothing forced the move: that stack is open source and we could have stayed on it indefinitely. We would rather build on the best tools for the job than stay on that stack out of loyalty — and a blockchain was never the part that kept those bookmarks private.

This one keeps the promise and rebuilds what it rests on. The encryption is now ours end to end: a key your password unlocks, with a recovery code as the second door. You create an account with a username. Nothing else.

Subscriptions. That is the whole model. There is no advertising, no data to sell even if we wanted to, and no investor waiting for us to change that.

It works because the architecture that keeps us out of your library also keeps it cheap to run: we store opaque blobs and do no work on their contents. That is why the free plan has no time limit on it. What Plus costs $48 a year pays for is the part that isn’t cheap: four clients, two servers, and the people who keep all of them working on platforms that change every year.

We intend to keep today’s free features free, and to never show an advertisement. If the app is worth paying for, some of you will pay for it.

A small team at STX Apps Co., Ltd. in Bangkok. Small enough that the person who answers support@bracemark.com is the person who wrote the code you are asking about — which is a support model with obvious limits and one real advantage.

Everything above is a claim about how the software works, made by the people who profit from you believing it. So the software is public: github.com/stxapps/bracemark-monorepo, under the AGPL-3.0. The apps, the browser extension, the two servers behind them, the key derivation, this page — one repository, all of it.

Being precise about what that buys you: you can read exactly what we do, and you can check that the encryption happens where we say it does. You cannot, from the source alone, prove that the code your browser just ran was built from it — that would take reproducible builds, which we don’t yet publish. Anyone who tells you otherwise about their own app is overselling it. What open source removes here is our ability to describe the design as one thing while shipping another.

One more absence, since you would find it anyway: no third party has audited this code. There is no report to link, no security firm on retainer, and no transparency report, because none of those exist yet rather than because they say something we would rather you missed. What is here instead is the repository, this page, and a policy that inventories every field we hold.