Privacy policy

Updated 28 September 2026

This policy describes the current Live Gossip Map website at livemap.heavenchester.com, operated by Heaven Lab. Contact: chester@heavenchester.com.

Information stored in your browser

Social links you import, titles, summaries, timestamps, location notes and engagement counts are stored in this browser's local storage, up to 100 imports. Your language preference is also stored locally. The app does not upload your saved social list to a shared application database. An optional, temporary Threads connection is separate from local saved imports.

Saved imports remain until you remove them, clear this site's browser data, or they are replaced by newer imports when the limit is reached. Installation and offline support cache an offline page and application icons; the service worker does not cache social feeds or government 3D models.

Requests needed to operate the map

Visiting the site sends network information, including your IP address and requested resource, to the hosting infrastructure. The application is hosted on AWS Lightsail in Tokyo. We use this infrastructure to serve the website and operate the Threads test connection. Your browser downloads 3D indexes, models and textures directly from the Lands Department using the configured browser-accessible government API key; Lightsail does not relay these files. Infrastructure and service diagnostics may process request metadata for operation and troubleshooting.

Map tiles, traffic information and place searches use third-party services, including OpenStreetMap and Hong Kong government map and traffic services. Place searches send location search text or the area being queried to the relevant service. Direct third-party requests disclose network information to that provider. The app does not request your device's GPS location.

Social previews

Map cards and selected report details can request provider-owned images or video posters. Selecting an Instagram report can load its captioned embed; an available Facebook platform preview opens when requested. A supplied playable video uses visible playback controls. The provider receives the requested post or media link and network information and may use its own cookies, particularly if you are logged in. Opening an original source also takes you to that provider. Their own policies apply: Meta Privacy Policy and Threads privacy information. You can use the map without opening embedded post previews.

When configured, the Threads test connection requests threads_basic and threads_keyword_search. Meta handles account authorization. Our server exchanges the authorization code and holds the access token, username and a session identifier in process memory for at most 55 minutes, until you disconnect, or until the server restarts. Expired sessions are removed within one minute. Tokens are not sent to frontend JavaScript or written to an application database. An essential HttpOnly, SameSite cookie identifies your connection; a separate ten-minute cookie binds the authorization attempt to this browser. Production cookies use HTTPS Secure protection. Search text is sent through our server to Meta using your own connection. Results are displayed in this browser and are saved locally only when you choose to add them. We do not persist search queries or results on the application server, and do not deliberately log credentials or authorization codes. Hosting providers may separately process request metadata. Two kinds of account can connect: nominated test accounts, and, only while we have explicitly switched on a private time-limited reviewer code, an app reviewer who enters that code at the reviewer access page. There is no general public sign-in, and the reviewer path is switched off unless we have enabled it for a review. A reviewer authorizes their own Threads account on Meta's own authorization page; we do not share our account, password or token with them, and their connection uses the same two permissions, the same 55-minute in-memory session and the same disconnection, expiry and deletion handling described above. The reviewer code is compared on the server, is not stored in your browser and is not written into any link; attempts are limited to bound guessing. Keyword search remains limited by Meta to the authenticated account's own posts pending review. This test connection is not used by the separate Gossip Scout public-web discovery process. It requests no posting or comment permissions.

Public-source discovery and the shared feed

Gossip Scout stores candidate public URLs, extracted public text, source timestamps, observed account names, approximate location evidence and review decisions on the operator's Mac mini. Accepted source observations are also stored in the private application database on Lightsail and exposed through the read-only shared feed. Parallel and Exa receive public search queries or source URLs for search/extraction. Source interpretation uses the operator's existing OpenAI connection through Hermes. This process does not receive visitors' browsing histories, private platform cookies or GPS locations.

Map-bubble thumbnails may request bounded public metadata through our server and redirect to provider-owned media. Media files are not downloaded into a permanent application archive or rehosted. Provider requests can disclose network information. The 24-hour display window and older-post fallback do not delete the underlying archive; records may be retained for review, deduplication and corrections. Contact the operator with a source URL for a correction or removal request.

Use, sharing and retention

We use the information described above to display requested content, remember local choices, resolve place clues and operate the service. The current application has no advertising or analytics SDK and no feature that sells your saved imports. Third-party source content remains with its provider; removing a local import does not delete the original post.

We do not maintain central visitor profiles. Separately, published public-source observations and the private discovery review archive are stored by the operator, as described below. Browser data is retained as described above. Infrastructure records are separate from browser storage and are subject to the hosting service's operation and retention settings; we do not promise immediate deletion of all infrastructure records when browser data is cleared.

Your choices and requests

See data deletion instructions to remove local imports or site storage. For access, correction or deletion requests concerning information you have sent directly to Heaven Lab, email us with enough information to identify the request. Do not send passwords or access tokens. We may need to verify that a request relates to you.

We will update this page if the service's data handling changes.