Local browser rescue

Keep locally what your browser has already received.

Triluna Document Rescue is being developed to preserve document bytes or pages already delivered to, or rendered in, the browser. It works within the viewer’s ordinary access boundary; it does not promise an original file when only a preview is available.

Developer preview downloads

Download the Chromium developer preview.

Chrome Standard and Chrome Advanced are unsigned packages for manual testing. Choose a target, check its published SHA-256 checksum, extract the archive, then load its folder through chrome://extensions with Developer mode enabled.

The Dropbox-enabled extension build is not distributed until its public Dropbox app key and stable extension redirect identity are configured. The source package is not a live Dropbox connection.

Local developer setup

Local developer outputs are available for review.

From the repository root, run cd triluna-document-rescue, then pnpm install and pnpm run build.
The local outputs are triluna-document-rescue/dist/chrome-standard/, triluna-document-rescue/dist/chrome-standard-dropbox/, triluna-document-rescue/dist/chrome-advanced/, and triluna-document-rescue/dist/safari-standard/.
To load a Chrome output, open chrome://extensions, enable Developer mode, choose Load unpacked, and select the extracted output directory. This manual installation does not make the packages signed browser distributions or Chrome Web Store releases. Safari Standard has no verified browser installation path.

To generate the optional Chrome Dropbox output, set TRILUNA_DROPBOX_APP_KEY to your public Dropbox app key before running pnpm run build. Connect Dropbox through its user OAuth flow, then choose Save to Dropbox to start an upload. Safari Dropbox support remains undecided and is not included in the Safari Standard output.

Capture modes

Standard and Chrome Advanced have different permission boundaries.

Standard: The Standard output is limited to PDFs already held in a same-context blob: URL and visible page material exposed by the viewer. Standard never refetches an HTTP document URL and cannot capture HTTP PDF response bytes. It can omit unloaded pages, content in cross-origin frames, or material the viewer does not currently expose. Safari Standard has no Advanced network capture.
Chrome Advanced: Chrome Advanced requires the debugger permission in its manifest. Chrome treats this as a high-access permission because it can attach debugging tools to a tab and inspect network activity; the permission is mandatory for this build and Chrome presents its permission prompt. Start capture explicitly in the selected tab. If the PDF is already loaded, confirm the reload control; the first PDF response observed after capture starts is selected, and only bytes streamed by the browser are retained. The capture observes out-of-process iframe (OOPIF) targets and detaches when it ends or is cancelled.

Chrome Standard + Dropbox adds an optional provider to Standard capture. It requires a build configured with a public app key and separate user OAuth authorization; an upload begins only after the user chooses Save to Dropbox. Choosing Connect also asks Chrome for optional access to Dropbox's API and upload origins; other builds do not request those sites. This route describes locally built source outputs, not a public release.

Provenance before export

A saved preview is not silently relabelled as its source.

ORIGINAL: The complete source bytes were received and are saved byte-for-byte.
RECONSTRUCTED: A new file is assembled from page assets, images, HTML, or SVG that the viewer already exposed.
CAPTURED: Rendered pixels are preserved from a page or canvas. Text, metadata, vectors, attachments, and annotations may be absent.

The class must be shown before export and recorded in an optional sidecar manifest. A rescued artifact retains the source document’s available geometry and appearance as a user-source file. It is outside the TDS page-output contract and is not restyled to look like a Triluna publication.

Privacy and temporary data

Document bodies stay outside Triluna.

A document body is not sent to Triluna. Capture data stays in browser-local temporary storage; the extension provides an Erase temporary data control and removes abandoned captures after 24 hours. Dropbox is separate: in the Chrome Standard + Dropbox output, the prepared artifact is sent directly from the browser to Dropbox only after user OAuth authorization and an explicit Save to Dropbox action.

Browser storage has finite quota; a large capture can stop when space runs out. Temporary rescue data expires after 24 hours without updates and can also be erased from the extension. Safari has reduced extension capabilities, and Safari Standard lacks Chrome Advanced’s streamed network capture. The Safari source output has not passed browser acceptance or Xcode packaging checks.

Dropbox is optional and currently built only for Chrome Standard. Safari availability remains conditional on a provider and browser architecture decision and has not been verified.

Lawful access

Preserve only what you may lawfully keep.

Use Document Rescue only for material you own, are authorized to access, or may lawfully preserve. It must not bypass sign-in, CAPTCHA, payment, membership, download quotas, geographic restrictions, DRM, or other access controls. It must not probe for hidden pages, private object IDs, or guessed endpoints.

For Baidu Pan or Quark Pan, use the normal viewer and its visible controls. Rescue may preserve only pages or resources the authorized viewer makes available; it does not use private endpoints or change account entitlements.