Compaction
Sign in

Privacy Policy

Version 1.0. This version applies from the date this page is published.

Who is responsible

The controller is Pohlmann Ventures UG (haftungsbeschränkt), registered with the Commercial Register of the Local Court of Charlottenburg (Berlin) under HRB 222538 B. Contact: hello@compaction.dev. Provider identification is on the Impressum.

The short version

  • Your prompts, your code, and your completions are never sent to us. Optimization runs on your machine.
  • The Compaction CLI works with no account at all, and in that mode it makes no calls to us.
  • If you create an account, the CLI does talk to our service — to activate the device, to get an entitlement, to check for Engine releases, and to record allowance usage. Those calls carry identifiers, counts, and status labels. They do not carry your content.
  • We run no analytics and no third-party tracking on this website.

This website

The website is served from Vercel’s hosting platform, which processes the connection data any web server receives, including your IP address, in order to deliver the page and to keep the service secure and available.

There are no analytics scripts, no advertising scripts, and no third-party trackers on this site. The only thing we store in your browser without asking is your light/dark theme choice, which is kept in local storage on your own device and never sent anywhere. If you sign in, a session cookie is set so that you stay signed in; that cookie is necessary for the sign-in to work.

Creating an account and signing in

Accounts and sign-in are handled using Supabase. You can sign in with GitHub or with an email address.

  • With GitHub, GitHub tells us the account identity you authorise it to share — normally your email address and your GitHub user id. We do not receive your GitHub password, and we do not read your repositories.
  • With email, we hold the email address you give us.

We use this to identify your account, to sign you in, and to contact you about it.

Activating a device

To use your account from the CLI you activate a device. Activation sends us a public key generated on that device and a name for it, and the browser page you approve it on shows you the device name, platform, and processor architecture so you can confirm you are approving the right machine. We store the device registration so that the entitlement can be issued to it and so that you can revoke it later.

The device token the CLI receives is a credential. It is stored on your machine with owner-only permissions and is sent only to the service address you activated against.

Entitlements and the Hybrid Engine

Your account is issued a signed entitlement, which the CLI checks locally before it starts the Engine. Requesting one sends your device token; the response is the signed entitlement, which is stored on your machine.

The CLI asks our service which Engine releases are published and downloads the signed artifact. That request identifies your device and the platform it needs a build for.

The Engine itself runs locally on your machine. The CLI starts it as a child process and talks to it over a local channel. Running the Engine does not by itself upload your prompts or your code to us. The Engine is never given your provider credentials.

Allowance and usage records

Community includes a monthly optimized-input allowance, so we have to count what it consumes. When a request is optimized on a metered route, the CLI records a usage event and sends it to us so the allowance stays accurate across your devices.

A usage event contains only:

  • opaque identifiers for the event, the local receipt, the entitlement, and the device;
  • the allowance period and the time the request was optimized;
  • labels describing the route type, the workflow, the provider name, and the meter version;
  • token counts.

It contains no prompt, no completion, and no code. This is what we mean when we say the records are content-free: it is a property of the record format, not a promise about our intentions.

The waitlist

If you join the Pro or Team waitlist we receive the email address you enter and which plan you were interested in. We use it to tell you when that plan is available. Write to us and we will remove you.

The CLI on your own machine

Receipts, logs, and configuration written by the CLI stay on your machine, under your project directory and your home directory. We do not collect them. If you send us a diagnostic bundle, you generate it yourself and choose to send it — nothing is uploaded automatically.

With no account, the CLI makes no calls to our service at all. Output shaping works in that mode and needs nothing from us.

Your AI provider

Compaction sits in front of the provider you chose. Your request still goes to that provider, using your own credentials, under your own agreement with them. We are not a party to that traffic and we do not receive a copy of it.

Email

If you write to hello@compaction.dev we receive your message and your email address and keep them so we can answer you.

How long we keep things

Account data, device registrations, and usage records are kept while your account exists. Ask us to delete your account and they go with it. Waitlist entries are kept until the plan launches or until you ask us to remove you.

Your rights

Under the GDPR you can ask for access to your personal data, correction, deletion, restriction, portability, and you can object to processing. Write to hello@compaction.dev and we will act on it. You also have the right to complain to a supervisory authority in the EU country where you live or work.

Changes

We will update this page when what we do changes, and the version number at the top changes with it.