Privacy

Information in your account

CaptureMem stores your account email address and identifier, the URLs you submit, and the library information associated with those links. That includes titles, source descriptions and readable source text, repository metadata, summaries, collections, tags, favorites, personal notes, and save or update timestamps. Supabase manages authentication credentials and password-recovery verification. We also record the version and time of account terms you accept, where the account flow asks for agreement.

If you import a list or a CaptureMem export, the submitted links and supported library fields are processed as part of that import. If you request audio or video processing, the source and media URLs, transcript, timestamped speech segments, sampled on-screen text and detected links, processing status, and credit records are associated with your account. Uploaded videos and generated video artifacts are queued for deletion 24 hours after their job ends; unclaimed uploads are cleaned up after their upload window expires plus 24 hours. Cleanup is periodic and may run later during an outage. Extracted text and provenance remain with your saved library.

For a saved public Instagram post, CaptureMem can read accessible slide images, keep extracted text and processing status, and save verified GitHub repositories as separate library links. Each discovered repository retains its source post and slide numbers. Public image addresses are kept in private processing records; downloaded image bytes are used during analysis and are not stored in your library. Existing saved repositories are reused. A video slide’s poster is an image; this automatic process does not watch the video or extract its audio.

For named-tool discovery, CaptureMem keeps bounded source-text snapshots, the names and quotations found in them, available slide numbers or timestamps, candidate public links, search citations, match scores, review choices and processing status. Saved matches retain their source history. Existing library entries are reused, including their archive or trash state. Private snapshots, processing credentials and carousel image addresses are not included in the browser’s discovery results.

Account and connection records include credential labels and scopes, creation and last-use dates, and access or revocation status. CaptureMem also keeps administrative audit records and, when used, billing, purchase, and credit records. An account-closure request adds its request time, status, and the operational records needed to cancel billing and complete deletion. Creator-program participation can add referral identifiers, commission records, a payout handle and payee name, and recorded payout references.

Public libraries and stars

Your library starts private. If you turn on public sharing, your chosen display name, custom public link, profile description, included folder names and descriptions, and the public source details for active links in those folders become available to anyone with the address and may be indexed by search engines. Public source details can include a source URL, name, domain, generated or source description, content type, repository language, license, and public GitHub star count.

Personal notes, “why saved” text, imported URL history, private processing records, inbox, favorites, archive, trash, and folders you mark private are not included on the public page. Folder settings inherit the main public switch, and you can exclude individual folders. Turning public sharing off removes the public page; changing the custom link makes the old address stop working.

Signed-in CaptureMem members can star a public library. We store the member and public-profile identifiers needed to prevent duplicate stars and show a total. Public pages do not show who starred a library. You can remove your star, and deleting either account removes its associated star record.

What the Chrome extension does

Opening the extension reads the active tab’s URL and title to show the page you are about to save. Selecting Save to Library, using its keyboard command, or retrying a failed save sends the selected URL to your configured CaptureMem site with your capture credential.

You can also manually paste a URL into the popup and choose Save pasted link. The extension does not request clipboard access or automatically read clipboard contents. An unsubmitted URL stays in the popup’s memory and disappears when it closes. Submitting sends only that URL to your configured CaptureMem site with your capture credential; its result and any failed URL use the same local save/retry storage described below.

The extension also provides Add from Tab Groups and Add from Bookmarks. Each opens a picker in a separate tab. Choosing Allow access requests the optional Chrome permissions for that source: tabs and tab groups, or bookmarks. With that permission, the picker reads open tab groups across your Chrome windows or your bookmark tree, including URLs, titles, group names, and folder names. This information lets you search and select individual links, groups, or folders locally. Closed saved tab groups must be opened in Chrome before they appear.

Nothing from that preview is sent to CaptureMem until you choose Import selected. The import sends only the selected URLs with your capture credential. Browser titles, folder names, group names, and the selection hierarchy stay in the picker; they are not sent to the server or recreated as library collections. CaptureMem may separately retrieve a title and other metadata from the saved source.

The extension does not modify bookmarks or tab groups, continuously collect your browsing history, inject scripts into visited pages, read page bodies, or request remote favicon images for the picker. Readable source content is retrieved by CaptureMem’s server after you save a link.

When you choose Connect to CaptureMem, a CaptureMem sign-in and confirmation page uses your existing website session. After you approve, a short-lived, single-use authorization code is exchanged for a credential that can save links. Your website session cookies and password are not copied into the extension. The extension receives your account email to show which library is connected. The connection expires after 90 days and can be revoked earlier.

The extension uses Chrome’s local extension storage for:

• Your configured CaptureMem site address, capture credential, its expiry, and the connected account email used to identify your connection.
• The most recent save result, including its URL, status, message, and timestamp.
• A failed URL waiting for a retry, when there is one.

These settings use local browser storage, not Chrome’s extension sync storage. The extension requests access to the CaptureMem site you configure so it can submit saves. Its local data supports connection setup, saving, status display, and retry.

The import preview, search, selections, and per-link results are held in the import tab’s memory, not saved to Chrome storage. Closing or reloading that tab discards them; links already saved remain in your library. Keep the tab open to view progress or retry remaining links. Stopping an import or changing the connected account prevents the next queued request; a request already sent may still finish in the original account.

Chrome Web Store Limited Use

CaptureMem’s use and transfer of information received through Chrome APIs complies with the Chrome Web Store User Data Policy, including its Limited Use requirements.

This information is used to save, organize, retrieve, and support the links you choose. Transfers to service providers support those functions, as described in this notice.

CaptureMem does not sell your saved links, notes, or information received through the extension to brands, advertisers, or data brokers. It does not use that information for unrelated advertising or creditworthiness decisions. There is no active brand-data sales program.

How saved content is processed

After saving, CaptureMem requests the URL from its source website or a repository service such as GitHub. It extracts accessible source information, prepares a summary, and matches topics to collections. The source website receives that request. Private or restricted content may be unavailable.

If no existing collection fits, CaptureMem may store a proposed broad folder and send an approval email containing the suggested name and relevant saved-source details. The signed-in review page lets you approve, rename, or dismiss the proposal. Opening the email or its link does not create the folder or file the saved item.

AI summaries use OpenAI only when an administrator has configured and enabled that connection. Without it, CaptureMem can use source-text summaries and filing rules. Audio processing requires an explicit transcription request and an approved minute limit. Short-video processing requires a separate credit confirmation and accepts at most 60 seconds and 100 MB. Saving a video link alone never starts paid media processing. Exact public GitHub references can be verified and saved automatically.

With the AI connection enabled, CaptureMem can also identify up to 15 software names in readable page text, post captions, carousel images or completed speech and on-screen text. It sends each name and up to 1,200 characters of supporting source quotations to OpenAI web search, then retrieves candidate public pages to check the matches. Saved personal notes and account metadata are not used for these lookups. Results offer up to three cited destinations per name. A verified primary match with an exact name and supporting citations can be saved automatically only at a score of at least 95/100 and a lead of at least 15 points over the next candidate. Other matches remain for your review. Scores describe evidence strength, not statistical probabilities. This process does not start audio or video extraction or spend media credits.

Services involved

CaptureMem uses these services for the purposes below. Their processing and retention also depend on their service terms and the settings of the relevant service account.

Supabase

Account authentication, sign-in sessions, and database storage for your library, access credentials, settings, and processing records. Google sign-in, when you choose it, supplies account identity information through Supabase.

Vercel

Hosting and running the CaptureMem website, APIs, and background processing. Requests to CaptureMem pass through this infrastructure.

Resend

Delivery of account invitation, confirmation, and password-recovery emails sent by Supabase, and transactional messages sent by CaptureMem, including proposed-folder approval emails. This involves the recipient email address and the message. A folder-approval email includes the proposed folder, relevant saved-source details, and a link to review it after sign-in; receiving or opening the email does not approve or file a link.

OpenAI

When an administrator enables AI and configures a key, readable source extracts and associated source information are sent to the selected OpenAI model. Saving a public Instagram post also sends up to 20 accessible slide images to that model to read visible text and exact repository references. Named-tool discovery uses extracted names and short source quotations with OpenAI’s web-search tool to find possible repository or app destinations, even when no URL appears in the source. Personal notes, edited description overrides, account identifiers, credentials and browser bookmark/group names are excluded from discovery input. Existing library folder names and descriptions are included for categorization, but not for named-tool lookup. Completed audio transcripts may be summarized. When selected for video extraction, OpenAI receives sampled frames and returns on-screen text and links.

Coconut and Deepgram

For media processing you explicitly request, Coconut receives the media URL and extracts audio, and for short videos, sampled frames. Deepgram receives the extracted audio for speech transcription. CaptureMem stores text, timestamps, job status, and credit usage. Short-video uploads and generated artifacts use private Supabase storage with temporary access links.

Google Gemini

When selected by an administrator for short-video extraction, Gemini receives sampled video frames to read on-screen text and links. CaptureMem stores the extracted text with timestamps and the model used.

Apify

When configured and you request processing of an Instagram post, the post URL is sent to Apify’s Instagram Reel Scraper to retrieve an accessible video. CaptureMem does not send your Instagram password or session cookies. Restricted or unavailable posts may require you to upload the video instead.

Stripe

When billing or credit purchases are available and you use them, Stripe handles checkout and payment details. CaptureMem sends an account reference and email address and keeps customer identifiers, subscription or purchase status, and accounting records.

When you connect ChatGPT or Claude

An MCP connection asks you to sign in and approve read access. The assistant you authorize can receive search results, saved source details, notes, collections, tags, completed transcripts, and video on-screen text with timestamps when it requests them. The current connection cannot add or edit links or start paid media processing.

The connected assistant’s own data handling applies to information it receives. Revoking a connection stops its future authorized access to CaptureMem; it does not erase information already included in an assistant conversation or stored by that service.

Cookies and browser preferences

The website uses cookies for sign-in sessions and connection security. It also stores your chosen library layout and guided-setup progress in your browser. Setup progress includes the connection type, step, completion acknowledgments, and an opaque credential identifier when needed; raw access credentials and test URLs are not persisted as setup progress. When an active creator referral link is used, a signed referral cookie can associate a later eligible purchase with that creator. Extension storage is separate from these website cookies and preferences.

Your controls and account deletion

• Manage connections. Revoke a capture or read-only credential in Settings & connections. Disconnect an assistant in that assistant’s settings as well.
• Clear the extension. Use Disconnect to revoke the current capture credential and clear local extension data. If the server cannot be reached, the extension explains that access still needs to be revoked in Settings & connections. Uninstalling clears local extension data but does not revoke server access or delete your saved library.
• Export your library. The library’s import dialog includes an Export JSON option for your links, notes, collections, tags, and memberships.
• Control public sharing. Use Share library in the dashboard to publish or unpublish your profile, change its address and description, and keep selected folders private.
• Understand Trash. Moving a link to Trash keeps it in your account so you can restore it. Archiving and trashing are not permanent erasure. Work already submitted for audio or video processing can still finish.
• Recover account access. Use Forgot password to request a recovery message. The page gives the same response whether an email has an account, to avoid exposing membership. Recovery verifies the emailed link before allowing a password change.
• Cancel a subscription. Open Manage billing to stop the next renewal. Ordinary subscription expiry preserves access to read and export the existing library; it does not delete the account.
• Close your account. Close account records a deletion request and disables account access and connections immediately after confirmation. CaptureMem cancels recurring billing, resolves any paid processing already underway, removes private uploaded media, and deletes the authentication account and saved library. The page explains the effect on paid access and unused credits before you confirm.

Saved library information remains stored when you sign out or uninstall the extension. Account deletion is processed as a tracked request rather than erasing financial or unfinished processing records immediately. Minimum billing, credit, and audit records, including internal account identifiers and Stripe references, may be retained for accounting, payment adjustments, security, and legal obligations. These records are not an active library or permission to market to you.

Copies in backups and records held by connected services may remain under their own retention controls. Deleting CaptureMem does not erase material already exported or copied into an assistant conversation. Contact us to request access, correction, or deletion of information, ask about a pending deletion, or exercise any privacy rights available where you live. We may need to verify that the request is yours.