Skip to main content
Orilami

Security Architecture

How Orilami is built to keep your files private: a local execution model, a zero-log architecture, and client-side verification you can check yourself.

Orilami's privacy isn't a policy you have to trust — it's an architecture you can verify. Every tool is designed so your files are processed on your own device and never uploaded. This page explains the three pillars that make that true.

How data moves when you use an Orilami toolYour file, the tool engine and the result all stay on your device. While you use a tool, only three kinds of small request can leave it: sign-in and token charges to Orilami (a tool name and a count, never a file), an optional 32-byte fingerprint to a timestamp authority, and card details typed on the payment provider's page when you buy tokens.Your deviceYour fileread into memory, never uploadedTool engineruns in your browserYour resultsaved straight to your deviceFiles and their names stay in this boxonly small requests:Orilami servertool name + count, sign-inTimestamp authority32-byte fingerprint, if you askPayment provideryour card, on their page
Solid lines stay on your device. Dashed lines are the only requests that leave it, and none of them carries a file.

Local Execution Model

All file processing happens in your browser, on your machine. When you open a tool and load a file, the bytes are read into memory locally and never sent to a server.

  • Heavy work runs in Web Workers so your file is handled off the main thread and the page stays responsive.
  • Documents, images, audio and video are read, rewritten and re-encoded entirely client-side.
  • Because there is no upload, there is no server-side copy to store, cache, log, or leak.

The strongest privacy guarantee is not "we delete your data" — it's "we never receive it."

Zero-Log Architecture

We can't log what we never see. Since your files are never transmitted, there are no file contents, no filenames, and no derived data sitting in server logs or analytics.

  • Your documents and images are not part of any request to our servers.
  • Account features (sign-in and your token balance), the optional timestamp relay and the support form are the only things that talk to our backend, and none of them carries the files you process.
  • Nothing about the content of your work is recorded on our side.

What never leaves your device, and what may use the internet

This list comes from the requests the site actually makes. If a tool ever needs a new kind of request, it is added here.

Never leaves your device

  • The files you open in a tool, and their names
  • Everything inside them: text, images, metadata, hidden content
  • The results the tools produce
  • What a tool found: detected names, numbers, faces, findings
  • Passwords you type to open a locked PDF
  • Key files, passphrases and licence signing keys
  • Buyer lists and other CSVs you load into a tool

May use the internet

Loading the site
Pages, scripts, fonts and the on-device models (face finder, text recognition, name detection) download from orilami.com, like any website.
Signing in
Your email and token balance live in our account database. Sign-in may show a Cloudflare check against bots, and a new account sends a hashed device check that stops free-token abuse.
Paying for a download
The tool's name, how many files or pages, and any report you add. Never the file.
Buying tokens
Checkout happens on Creem's page; your card details go to them, not to us.
Trusted timestamps (optional)
Contract Date Stamper, Chat Evidence Archiver and Digital Shredder can send a 32-byte fingerprint and a random number, through our relay, to a timestamp authority.
Bitcoin anchor (optional)
Contract Date Stamper sends a blinded fingerprint to the OpenTimestamps calendars, and verifying it reads public block data.
Writing to support
What you type in the support form, sent to our support inbox.

Client-Side Verification

You don't have to take any of this on faith. Because everything happens in the browser, you can confirm it yourself:

Verifiability is the point. A promise you can check is worth more than one you can't.

How it's built, honestly

We'd rather tell you exactly how the tools work than hide behind buzzwords:

  • Plain client-side JavaScript and Web Workers. The processing runs as JavaScript in your browser, with the heavy work moved into Web Workers. We don't claim exotic technology we don't use.
  • A strict content security policy. The site tells your browser which servers it is allowed to talk to. This limits where any data could go, and you can read the policy in the response headers yourself.
  • The paywall is honest about what it can do. When a tool costs tokens, the charge is checked on our server by deducting from your balance — not by trying to lock the file inside your browser. A file your browser can open is a file you can save, and we won't pretend otherwise. What you pay for is the processing, not a lock we can't really enforce.

Why the work stays in your browser

A purely client-side tool cannot stop a determined user from reading the processed bytes: anything the browser holds sits in memory, reachable through DevTools or a heap snapshot, and a key shipped to the client can be extracted. Access control only works at a server boundary, so Orilami does not pretend otherwise with a fake in-browser lock. That is why the token charge happens on our server when you download, and why the files themselves never need to leave your device.

Where accounts fit in

Orilami has optional accounts for your token balance so it follows you across devices. Signing in and topping up communicate with a backend — but your files never do. The two are kept strictly separate: the tools work the same whether or not you're signed in, and processing always stays on your device.

When a tool charges tokens, that charge is recorded: which tool, which account, how many tokens, whether it went through, and when. We keep it to spot tools that are failing and to put tokens back when they do. It describes the transaction, never the file — what you processed, what the tool found, and what it produced stay on your device and are never sent anywhere.

To be concrete about what does leave your device when you use an account:

  • Only your sign-in and token balance. Accounts run on a managed database provider. The backend stores your email and your token count — nothing about the files you process.
  • The files never travel with it. Loading a file, processing it, and downloading the result are entirely separate from any account request. One does not touch the other.
  • You can use the tools signed out. Accounts exist for billing and syncing your balance, not for the tools to work.

Who this matters for

Keeping files on your own device isn't a nice-to-have for everyone — for some work it's a duty:

  • Lawyers and notaries handling case files, contracts and client records under confidentiality rules.
  • Doctors and clinics working with patient documents and photos they can't upload to a random service.
  • Accountants and finance teams with statements, tax files and account numbers.
  • Journalists and their sources who can't risk a sensitive document touching a third-party server.
  • HR and recruiters processing resumes, IDs and personal data for many people.
  • Government and legal filing where ID photos and documents must stay private while being prepared.

For all of them, "the file never left my device" isn't marketing — it's the reason the tool is safe to use at all.

Responsible disclosure

If you think you have found a security problem in Orilami, please tell us first and give us time to fix it before you share it. Write to security@orilami.com with what you found, the page or tool, and the steps to see it happen.

In scope

  • orilami.com and its pages, tools and API routes.
  • Anything that would let a file, its name or its contents leave your device without you choosing it.
  • Ways to take tokens, change a balance or reach another person's account.

Please don't

  • Run denial-of-service tests, flood forms or send spam.
  • Open, change or keep data that belongs to other people. Use your own account for testing.
  • Use social engineering or physical attacks, or test the services we rely on (sign-in, payments, email).

What you can expect

  • A first reply within 48 hours on working days (Monday to Friday).
  • Updates while we work on it, and a note when the fix is live.
  • We will not take legal action against research done in good faith that follows these rules, and we will credit you if you want to be named.

Want to see it in action? Start with the Metadata Remover or the Sensitive Info Remover tool, or browse all tools.