Privacy Policy

1. Two kinds of data

• Data about you: account, sign-in, billing, usage and support data. We decide how it is used.
• Customer Content: the code, files, data, secrets and output that you or your agents put on boxes and sandboxes. We process it for you to run the Services, under our Terms. If it contains other people's personal data, you are responsible for having the right to process it.

2. What we collect

Account and workspace

Your name, email address, profile picture and provider account ID from Google or GitHub, or the email address you verify with a code (terminal or AI-agent sign-up). Workspace details: name, icon, members and roles, claimed domain, invites, settings, policies, and API key names, including the name an agent gives itself.

Sign-in records

The time, network (IP) address and browser of each sign-in, to investigate abuse and account takeover. You can see yours under Settings → Security. We also keep a record of which Google or GitHub account signed in as which user, so a recycled mailbox can't take over a deleted account. When the free credit is claimed through agent sign-up, we keep a one-way hash of the email address so the same inbox can't claim it twice.

Billing

Stripe processes payments and holds your card and billing details. We store the plan, Stripe's references to your customer record, subscription and payments, payment amounts and status, and a credit ledger and hourly usage for each workspace. We never see your full card number.

Website and dashboard analytics (PostHog, US cloud)

• Page views, clicks and named product events: sign-up and sign-in provider, box actions (create, pause, resume, resize, delete, auto-pause), opening the browser terminal, signing in from the CLI, copying a command, onboarding steps, starting checkout (with the plan amount), activating a plan, and scroll and video progress on campaign pages. Events note whether the page was opened in a social app's built-in browser.
• After you sign in, events are linked to your account, including pages you viewed earlier in the same browser.
• Before an event leaves your browser, its URL is cut to the path: query strings and fragments are dropped, and tokens in private links are replaced.
• Clicks are captured automatically. A click in a box's Files tab can record the name and path of the file clicked.
• Errors in your browser are sent to PostHog and linked to your account's name and email.

Session replay (PostHog)

While you are signed in, the dashboard records a replay of the interface. Every input and all text are masked in your browser before anything is sent, so a replay shows layout, clicks and navigation, not the words on screen. The terminal, desktop view, secrets panel and connector form are recorded as empty placeholders. Signed-out visitors are never recorded, so the marketing pages and docs produce no replays.

Vercel Web Analytics

Aggregate page-view counts by page, country and browser. It isn't told who you are and sets no cookie.

Cookies and browser storage

• Sign-in session cookies, which the dashboard needs to work.
• prized_src, a first-party cookie kept 30 days and set on the first page you land on. It holds the referring site's hostname, the landing path (tokens removed) and campaign tags (utm_source, utm_medium, utm_campaign, ref). It contains no identifier.
• A PostHog identifier that links your events, kept in a first-party cookie and in your browser's local storage.
• Shared port links set a viewer cookie on the link's own domain, valid up to 12 hours.
• The launch film on the home page plays from YouTube's privacy-enhanced domain (youtube-nocookie.com) when you open it; nothing is sent to YouTube before then.
• We don't use advertising cookies or pixels.
• The New box form and onboarding suggest the nearest region using an approximate location that our host derives from your IP address on each request. It isn't stored. Your browser also times a few empty requests to each region's AWS endpoint; the results stay in your browser.

CLI telemetry

While you are signed in, the CLI records which command ran, its version, your operating system, how long it took and its exit code, and sends counts in batches. A failed prized setup adds one word naming the failed check. It never records arguments, flags, paths or file contents, and sends nothing while you are signed out. The CLI and the box agent have no crash reporting. To turn telemetry off, set enabled = false under [telemetry] in ~/.config/prized/config.toml, or set PRIZED_NO_TELEMETRY=1.

Box and sandbox operations

• Metrics: CPU, memory, disk and network use.
• Process and port snapshots: listening ports and the busiest processes, including each process's command line (shortened), which can contain arguments. This is also how we detect prohibited use such as mining.
• Command history: for commands run with prized exec or the API, we keep the command (first 8,192 characters), directory, who ran it, when, how it ended, and up to the last 1 MiB of each output stream. Members who can see the box can read it.
• Audit log: workspace actions such as lifecycle changes, settings, keys and certificates, secrets, membership and policies, plus commands, file transfers and prompts that the CLI or box reports. It can contain a command line, a path or a prompt's first line, and so any token typed into one.
• Secret usage: per secret and host, how often and when a value was used. Never the value, path, header or body.
• Secrets: stored encrypted. Our servers hold the encryption key and decrypt a secret only to deliver it to the boxes it is mounted on, and a sandbox's environment variables only to deliver them to that sandbox. A value can't be read back through the dashboard or the API.
• Sessions in transit: browser-terminal, file, desktop, phone and shared-port sessions pass through our relays, which can read them in transit but don't store them. The fallback relay for networks that block the tunnel carries only encrypted traffic it can't read.

We don't collect code, file contents, terminal output, shell history, secret values or CLI arguments for analytics.

AI advisor (optional)

If you use the environment advisor in prized setup, an inventory of your machine (tool names, versions and sizes) and your project's manifest files are sent to Anthropic to produce suggestions, with secret values removed. Nothing is sent if you don't use it. AI tools you run on your boxes, such as Claude Code or Codex, talk to their providers under your own accounts, not ours.

Support and email

Messages you send us, and our replies, are kept in our company email (Google Workspace). We send account, billing and security email through Cloudflare. We may also send a welcome note after you sign up and a reminder or two if you haven't started a box, and the founders may email you directly about your account or to ask for feedback. We don't send newsletters.

Logs and error reports

Our servers keep logs at AWS and send server-side error reports to Sentry. A log line or error report can include an email address, a token or other data from a failed request. We work to scrub secrets but can't guarantee it.

3. How we use it

• Provide, bill for and support the Services.
• Keep them secure: detect fraud, trial abuse, mining and other AUP violations, and investigate account takeover.
• Understand and improve the product, using analytics, session replay, telemetry and error reports.
• Send service, billing and security notices, and reply to you.
• Comply with law and enforce our Terms.

We don't use Customer Content to train AI models. We don't sell personal information.

If you are in the EEA or the UK, we rely on these legal bases: performing our contract with you; our legitimate interests in securing, understanding and improving the Services; legal obligations, such as keeping tax records; and your consent where the law requires it.

4. Who we share it with

We don't sell personal information or share it for cross-context behavioral advertising. These sub-processors handle it for us:

We also share data with people in your workspace as their role allows (for example, members can read command history and the audit log), with an endpoint you set up for audit export, with professional advisers, with authorities when the law requires it, and with a buyer in a merger or acquisition. When this list changes, we will update this page and email workspace owners.

Staff access. Authorized Prized staff can access production systems, including the hosts that run boxes and sandboxes and the control plane, for operations, security, support and abuse investigations. That access can reach Customer Content.

5. Where data is stored

• A box's disk, snapshots and daily disk backups stay in the AWS region you pick (Oregon, Northern California, Northern Virginia, Ohio or Frankfurt) unless you move the box.
• These are stored in the US, whatever your box's region: account and billing data, the control-plane database (including command history, audit log and metrics), analytics, server logs and database backups. Browser-terminal, phone and shared-port sessions are relayed through Oregon.
• Sandboxes run only in Oregon (us-west-2).
• Cloudflare, GitHub and Google may process data outside the US.

When we transfer personal data from the EEA, the UK or Switzerland to the US, we rely on the European Commission's Standard Contractual Clauses and their UK and Swiss equivalents.

6. How long we keep it

After a workspace is erased (with Erase now, or 30 days after closing), we no longer hold its boxes, disks, snapshots, secrets, environments, tokens, members, ledger, usage, command history, audit log or policies. These remain for the periods above: invoices and payment records at Stripe; sign-in identity history; sign-in records; the free-credit email hash; other members' accounts; PostHog data, Sentry reports and server logs; backups; and fleet-wide aggregates.

7. Your choices

• Do Not Track. We honour it. With DNT on, the site records no page views, events or replays, doesn't set prized_src and removes an existing one. One exception: when a plan payment clears, our server can record that the plan was activated.
• Ad blockers may block our analytics; the dashboard still works.
• CLI telemetry can be turned off (Section 2).
• Zero data retention, on custom plans, shortens metric retention, stops process history and command content, and stops final snapshots of deleted boxes.
• Deletion. You can delete boxes, snapshots and command history records yourself, and the owner can close or erase the workspace.
• Email. Service email is part of the Services. To stop welcome, reminder or founder emails, reply to one and ask.

8. Your rights

Depending on where you live, you may have the right to access, correct, delete or export your personal information, to object to or restrict some uses, and to withdraw consent. Email hudson@prized.dev. We will verify your identity and reply within one month. We won't treat you differently for using these rights. If you are in the EEA or UK, you can also complain to your data protection authority.

Some data we must keep, such as tax records. Backups age out on the schedule above and aren't edited one person at a time. When you ask us to delete your data, we also delete your analytics profile and, where we can find them, error reports and logs about you. If your data is in someone else's workspace as Customer Content, contact that workspace's owner.

9. Security

Box disks and snapshots are encrypted at rest. The site and API use HTTPS only. Box access uses short-lived SSH certificates. Secrets are encrypted at rest. A SOC 2 audit is in progress; there is no report yet. No system is perfectly secure.

10. Children

The Services are not for anyone under 18, and we don't knowingly collect children's personal information.

11. Changes

We will post changes here with a new date, and email workspace owners at least 30 days before a material change takes effect.

12. Contact

Prized Inc., [to confirm: mailing address]. hudson@prized.dev