Last updated:

What the Chrome extension handles

Keep NI collects and processes data to check the pages you configure and turn selected content into saved readings. This is data handling even when the information stays on your computer. The current version has no Keep NI online account or cloud sync.

For manual checks, previews, and scheduled checks, the extension opens a dedicated tab in the Chrome profile you select. It handles:

  • Full-page screenshots and structured page content, including text, accessibility labels, element states, and layout information.
  • Monitored URLs, the page URL after navigation, capture timestamps, snapshot identifiers, and capture dimensions.
  • A randomly generated profile identifier and a default or user-entered profile name, used to identify and select the Chrome connection in the desktop app.
  • Temporary session cleanup metadata: the dedicated tab identifier, session identifier, and session expiry time.

Capture is limited to the tab created for that session. The extension does not scan your other tabs or read Chrome’s browsing-history database. Opening a monitored page makes ordinary website requests through Chrome; that website handles those requests under its own practices.

From capture to AI processing

  1. The extension captures full-page evidence in the dedicated Chrome tab.
  2. It transfers the capture to the desktop app through Chrome Native Messaging and a local helper on the same computer.
  3. The desktop app keeps the capture in memory for display and processing. It saves monitor settings and completed check results locally, as described below.
  4. When a Read step runs, the desktop app sends the selected screenshot region, relevant structured page evidence, and field names, types, and instructions to the configured AI provider or model server. The response supplies the requested readings.

The full-page capture and the AI request are different. The desktop app crops the screenshot and filters the structured evidence for the selected region before sending a Read request. That evidence can include relevant accessibility labels and surrounding structural context. The full page is not automatically sent to the model; selecting the whole page can, however, include the whole screenshot. Region selection does not limit the initial full-page capture or remove sensitive information from it.

A remote AI provider or model server receives the request content. A model server can run on your computer or another machine; configuring a server does not by itself make processing local. The app also contacts the selected service to discover available models. The current user-configured server connection accepts HTTP or HTTPS; HTTP does not encrypt that network traffic. Review the service’s data practices and connection settings before sending content. Keep NI cannot control its retention, logging, or other processing and does not delete its copies when you delete local data.

Website sign-ins and sensitive content

The dedicated tab uses the selected Chrome profile’s existing website sessions. Keep NI does not read Chrome’s password store or export its cookies to the desktop workspace or AI requests. Chrome manages those sign-ins; the extension does not have a separate website login store.

Signed-in pages, URLs, screenshots, and page text can still contain personal or sensitive information, including names, messages, account details, financial or health information, or a secret displayed on a page. Structured extraction excludes password input elements, but this does not redact screenshot pixels or guarantee that secrets shown elsewhere are excluded. Choose monitored pages and Read regions with this in mind.

Storage, retention, and deletion

In the extension: screenshots and page content are handled in memory for capture and transfer, not saved in extension storage as a capture archive. Cleanup metadata lives in Chrome’s session storage and is removed after successful tab cleanup. Cleanup is attempted when a session ends, is stopped, expires, or loses its connection; failed cleanup can leave a tab or record until recovery succeeds. Chrome clears session storage when the extension reloads or the browser restarts.

The profile identifier and name persist in Chrome’s local extension storage between sessions. You can change the name in the extension popup. Removing the extension clears its extension storage, but does not delete data already transferred to the desktop app. Chrome’s history, cache, cookies, and website sessions are managed separately through Chrome’s controls.

In the desktop app: the local workspace database stores monitors, configured URLs, steps and field instructions, schedules, readings from completed checks, check history, model settings, and the provider API key. The app also saves the selected Chrome profile identifier in local preferences. Saved workspace data has no automatic time-based expiry.

Screenshots and structured page content, including page trees, stay in memory during an app session and are not saved in the workspace database. Restarting the app clears these captures. Test step previews and their readings are also temporary: saving monitor settings does not save test output. Manual and scheduled checks still save their results and execution details in History.

  • Delete a check in History to remove that check and its saved readings.
  • Delete a monitor to remove its saved workflow settings, schedule, checks, and readings. Shared model settings and the API key remain.
  • Use Remove key in the provider settings to remove the stored API key. This does not revoke the key with its provider.

Removing the desktop app does not automatically remove its workspace, preferences, development logs, or Chrome Native Messaging registration. These require separate local cleanup; there is no all-data removal control in the app. Database deletion is not a secure-erasure guarantee and does not remove backups, separate debug logs, Chrome data, or copies held by a provider. Contact support for help locating local data.

Purpose-limited use and sharing

Keep NI’s use and sharing of information received through Chrome APIs comply with the Chrome Web Store User Data Policy, including its Limited Use requirements. The purpose is to capture the pages you configure and produce and retain the readings you request.

Capture data passes to the local desktop app, and the content needed for Read steps passes to your configured AI provider or model server. Keep NI has no developer-operated capture upload, analytics, or advertising service. We do not sell this data, transfer it to data brokers, use or share it for advertising, or use or share it to determine creditworthiness or for lending. These limits also apply to readings derived from captures. We do not receive your local captures for human review; if you choose to share specific content with support, it is used to help with that request.

App downloads and update checks

Installers and the update feed are hosted on Google Cloud Storage. Release versions of the Mac app use Sparkle to check for updates daily; you can turn automatic checks off in Settings or check manually. Requests involve ordinary network information such as your IP address, requested URL, and app and system information in the request headers. Keep NI does not enable Sparkle’s optional system profiling or send monitor data, captures, readings, or API keys with update checks. You choose when to install an update and restart the app.

This website and support

This website is hosted on Google Firebase App Hosting. Serving the site involves ordinary request information, such as IP addresses, requested URLs, and browser request details, which the hosting infrastructure may process in access or operational logs. The site has no sign-in, data submission forms, or advertising trackers. Documentation search runs in your browser. The site does not connect to your local workspace or receive your monitor readings.

When enabled on the production website, PostHog measures pageviews, clicks on the Mac download link, and clicks on the Chrome Web Store link. Events include the page path, referring domain, basic browser and device information, and the Mac release version when downloaded. We exclude URL query strings, fragments, search text, and page contents. Click counts do not confirm a completed download or installation.

We use PostHog’s cookieless mode: it does not store analytics cookies or identifiers in your browser’s local or session storage. PostHog processes your IP address and browser user agent to calculate a daily hash for approximate visitor counts, then strips the IP address and raw user agent from these events. We do not identify you, create person profiles, record sessions, or track returning visitors across days. Website analytics is processed by PostHog in its US Cloud region. This does not add analytics to the Mac app or Chrome extension. See PostHog’s privacy policy for information about its processing.

If you email support@keepni.app, support mail is hosted by Namecheap Private Email. Your email service and the services delivering and receiving support email process your address, message, and any attachments. We receive what you send to respond to your request. No automatic retention policy has been configured for the support mailbox. Include only the information needed for support; do not send API keys or unrelated sensitive captures. You can use the same address for privacy questions or to request deletion of support correspondence. Deleting app data does not delete email correspondence or hosting logs.

See Set up AI models for current configuration details. We will update this page as the product’s data handling changes.