Last updated 25 October 2026 · prepared for personal use

Privacy notice

Cargo Cult is a private personal application operated by Jayson Winters. This notice explains how its own application code handles connected information.

Information accessed

Google sign-in provides a verified account identity and email to restrict entry to the owner. The River connection reads a designated Google Sheet containing the owner's catalogue and its operating context.

The optional Google Calendar connection requests read-only access to calendar events and the account's calendar list, plus account identity for verification. It displays event titles, start/end dates or times, location, calendar name, event links, and limited attendance/status information from two selected calendars over a seven-day window. Existing calendar sharing controls which sources can be read.

The optional Todoist connection requests read-only access to application data. It displays open tasks, descriptions, dates, deadlines, priorities, labels and projects. The owner can separately upload or replace catalogue images.

The optional Read and summarise action fetches a public web page and sends its bounded readable text, your entered comments and classification labels to OpenAI to draft a summary and suggestions. It runs only when you press that button. It does not send the private River Catalogue, tasks, calendar events or uploaded photos. You review generated text before saving it on Shore.

Why information is used

Information is used to authenticate the owner and display the requested personal workspace. The Calendar integration does not create, edit, delete or respond to events. The Todoist integration does not create, edit or complete tasks. The River display uses read-only access. Quick capture stores an arrival privately on Shore, separate from the River Sheet. Keep in River uses a separate server write client to append a Catalogue entry only after the owner reviews it and explicitly chooses to preserve it; it does not edit existing entries.

Storage and security

The application runs on SiteGround over HTTPS. Connection credentials and OAuth refresh/access tokens are stored in private server files outside the web-served and deployment directories. Google ID tokens used for the Calendar connection are checked and discarded. Remembered sign-in uses a signed, Secure, HttpOnly cookie with an absolute expiry of up to 30 days.

Task and event responses use short-lived server-memory caches and browser memory while the page is open. They are not saved by the application into browser localStorage or an offline service-worker cache. Unsubmitted capture drafts remain in browser memory while the page is open; reloading or closing the page loses an unsaved draft. Saved Shore arrivals, photos and transfer state are stored on the private server outside web and deployment directories. Let go hides an arrival in a recoverable view; it does not erase its files. Kept arrivals retain their transfer receipt and source photo for recovery. A private server capture journal retains submitted draft fields and the verified saved entry to prevent duplicate writes and reconcile interrupted saves. These journals remain until separately removed from server storage. A private server AI-usage ledger retains request fingerprints, conservative cost reservations and generated-result receipts so interrupted requests are not submitted twice. These receipts can remain even if you do not save the draft to Shore. Full page HTML is not retained. Summary requests set store:false; OpenAI may retain data under its own API data-handling policies. Theme preferences may be remembered on the device. Uploaded images and their metadata are stored separately on the server; previous image versions may remain for recovery.

Sharing and other services

Google and Todoist receive the API requests needed for their integrations, and SiteGround processes the hosted application traffic and server storage. Hosting infrastructure can retain operational logs and backups under its own policies. When the owner opens an external source link or asks the application to look up a public image source, that destination receives the request.

Cargo Cult does not sell connected data or use it for advertising. Its normal display and refresh code does not send Calendar content to an AI service or use that content to train AI models. The owner may choose to share information separately, including when seeking support. The conversation handoff copies navigation context and the question the owner writes; it does not automatically copy catalogue entries, tasks or events.

Cargo Cult's use and transfer of information received from Google APIs follows the Google API Services User Data Policy, including its Limited Use requirements.

Control, retention and removal

The owner can disconnect Calendar or Todoist from its Connection options. This clears the app's active saved connection and in-memory cache; it does not delete source records or change calendar sharing. Access can also be revoked in the provider's account settings. Signing out clears the app session; source records remain in their original services. Uploaded images and previous versions remain until separately removed from server storage.

For removal of application-held files, questions or corrections, contact jaysonwinters@mac.com. Operational backups or provider records may follow their provider's retention process. This notice will be updated when the app's data handling materially changes.