How to Preserve and Share Web Research Without Exposing Your Private Notes

A saved URL is a useful pointer, not a guarantee. Pages move, text changes, and access conditions shift. Good web research therefore separates the address, the evidence you retained, your private interpretation, and anything you intentionally publish.

Keep the strongest evidence Arivu can safely recover

Capture the page through the normal workflow. Safe direct extraction begins immediately and remains available on every installation. When complete browser capture is enabled, Arivu can also render the page in an isolated background service and use the rendered result when it provides stronger evidence.

That is evidence selection, not a race in which the last response wins. A challenge page, partial browser result, or failed attempt cannot replace a stronger direct reader or the last good copy. Capture history records what each attempt recovered, while provenance and quality details explain which evidence became active.

Selected passages are especially useful when a particular claim matters. Add your interpretation as a private annotation or Note, rather than blending it into the source.

Preserve the article, not only its text

Article figures often carry evidence that text extraction alone cannot keep. Arivu downloads reader images through the same bounded, safe fetching path, stores them under the bookmark owner’s account, and rewrites the reader to use those local copies. This works for ordinary direct captures as well as rendered captures. Safe MathML also remains available for equations, matrices, fractions, and other mathematical notation.

The reader’s preserved-copies panel can include several records with different jobs:

• a screenshot keeps visible layout;
• a PDF keeps a paginated rendered document when the operator enables it;
• a source response keeps what direct retrieval returned;
• self-contained offline HTML keeps a downloadable page copy.

Offline HTML is transformed into an inert, sandboxed preview inside Arivu. Pages above the preview ceiling remain downloadable instead of being loaded into an unsafe or memory-heavy browser view.

None of these formats automatically makes evidence authoritative, complete, or legally reusable. Complete capture does not receive your browser cookies and is not a way around authentication, paywalls, or publisher controls.

Let the installer own the browser boundary

Fresh Arivu installs on supported Linux hosts enable complete browser capture by default after showing the installation plan. The service is headless and separate from the core Go application. It does not open a visible tab, inspect your browser profile, or require you to install Docker, npm, Node, Chromium, or Monolith yourself.

The installer downloads and verifies the matching native runtime, checks it before activation, and manages upgrades, rollback, disabling, and removal. Screenshot and offline HTML preservation are enabled for fresh installs. PDF is disabled by default and requires separate operator configuration because of its additional storage cost.

Choose the core-only option when you want the smallest runtime or use a host outside the supported capture baseline. Direct extraction, locally retained reader images, capture history, and the last-good evidence rules continue without the browser service. Disabling complete capture later does not delete reader content or artifacts you already kept.

Separate private interpretation from public evidence

Private tags, Notes, annotations, backlinks, graph connections, and Insights often contain unfinished thinking. Sharing a live bookmark view can expose more than the source you meant to discuss.

Instead, create a public secret-link snapshot. Arivu copies selected public fields, sanitized reader content, and active reader media at publication time. The bookmark disclosure creates an item-only link. Authenticated share APIs and local automation can explicitly add a screenshot or PDF, while private workspace fields remain excluded. Raw source responses and offline HTML do not become public artifacts.

Treat the snapshot as a publication

Later edits, recapture, or deletion of the private bookmark do not rewrite the published item copy. An authenticated update can deliberately replace membership under the same secret link, so treat that operation as a new publication rather than an invisible sync.

Set an expiry when access should end automatically. Revoke or delete a share when it should no longer be available. Anyone holding an active secret link can read it, so do not use secrecy of the URL as a substitute for judgment.

A share can expose a bounded public RSS feed of its newest items. This is deliberate publishing, not social networking, collaborative editing, or a mirror of your private workspace.

Back up the whole preservation boundary

Self-hosting keeps the application and storage under infrastructure you operate, but it also makes recovery your responsibility. When retained artifacts matter, a database file alone is not enough.

Installer-managed backups pair a consistent SQLite snapshot with the adjacent asset store and write a versioned size/hash manifest that restore verifies. Full JSON remains the portable account-level export and restore format. Test the recovery path that matches your goal.

The durable pattern is simple: retain the strongest safe source evidence, keep interpretation private by default, inspect preserved copies according to what each format proves, and publish only an explicit snapshot you are prepared to revoke or let expire.

Continue with Capture and Library, Preservation and Sharing, and Import and Export.