04 / GOOD TO KNOW

FAQ

Getting started

Is SanFrame free, and do I need an account?

SanFrame is free to download and use, with no SanFrame account required. The built-in AI assistant needs your own DeepSeek API key. Conversations and connection tests make real requests to its online service and may incur charges. External AI apps and OCR model services may also charge separately; fees depend on the provider.

Which devices can I use?

An x64 installer is available for Windows. The macOS app requires macOS 13.4 or later and supports both Apple Silicon and Intel Macs. On Android 10 or later, install SanFrame Transfer to exchange text, images and files with a computer. Screen capture, OCR, PDF tools and content groups are desktop features for Windows and macOS.

Can I use SanFrame offline?

Capture, annotation, record management and built-in OCR work locally without internet access. The built-in AI assistant connects to DeepSeek online and needs internet access, as do external online AI services. Local OCR models need their model files and service prepared first. The optional privacy model must also be downloaded and enabled first. Device transfers do not need internet access, but the devices must share a local network and run their apps.

Where should I start?

Start by saving an image or a piece of text. Use Image tools and built-in OCR for images; edit, tag or search text in Pasted content. For ongoing collection, enable Clipboard auto-collection from the top toolbar, then choose the content types. It starts off. For software questions, configure the built-in AI assistant and choose Guide only. Model OCR, AI organization and device connections can be configured when needed. In Settings → General, choose the default workspace. Appearance offers five accent colors with Liquid glass or Classic styling and Light, Dark or System modes. Glass is rendered inside the app; it does not reveal the desktop. Switch workspaces automatically starts off. Enable it to follow the actual destination of new content, optionally excluding background additions. Switching waits while settings, PDFs or protected editing are open. It does not bring the window forward or enable collection or device connections. History, favorites and archives share the Unified list / By type preference. Use the switch beside navigation without changing where records are saved or clearing drafts. Drag navigation entries to reorder them; shortcuts follow the visible order.

What gets sent to other devices or services?

Collection and OCR

How do I automatically save what I copy?

Open the collection control in the top toolbar, enable Automatic collection, then choose text, images or files. It is off by default and only collects new copies while SanFrame is running. Text goes to Pasted content, image content to Image tools, and files copied in a file manager to File references. Collection does not run OCR on images or back up referenced files automatically. You can separately enable the Toggle auto-collection shortcut. Turning collection off or quitting keeps existing records, but content copied during that time is not collected later. Rapidly overwritten clipboard content may be missed.

How do captures and copied content get app tags?

With source identification enabled, new captures and enabled clipboard auto-collection can add tags by source app. Apps without assigned tags use their names by default, reusing an existing ordinary tag with the same name. Assign your own tags in Manage tags → App rules to take priority. An app can have multiple tags and apps can share tags. Turn app-name tagging off separately without removing existing tags. New captures go to Image tools without requiring clipboard auto-collection. Frozen and scrolling captures use visible windows in the selected region to identify the main app. System captures only have an inferred foreground app, which may differ from the selected region. Ordinary image imports, unclear regions and old records do not get guessed sources or retroactive tags. Enable text, image or file auto-collection separately and keep SanFrame running. Text goes to Pasted content and images to Image tools. Clipboard sources known only as foreground candidates require a separate per-app opt-in for inferred matching. Source identification starts enabled and preserves an existing off setting. It cannot identify every source or classify by meaning, individual web pages or chat contacts. Exclude apps you do not want to auto-collect; this does not block explicit captures or remove old material. Hiding a source only removes it from the default list; restore it under Hidden. Collection and rules stay active. Source records and rules stay local, while ordinary tags containing app names follow your sync and AI access settings.

How do I save Handy transcripts?
Will repeated captures or imports create duplicate records?

Valid identical records are usually reused, keeping tags, favorites, archive state and OCR. Active reimports update the recent position without changing original dates or existing groups. Archived records are not automatically restored. Duplicate image check defaults to Fast, comparing identical image files. Thorough can also match exactly identical pictures saved by different apps. Merely similar images remain separate. Save identical screenshots once starts on and can be disabled separately without disabling deduplication for imports or clipboard images. Text comparison preserves whitespace and case. File references compare original path, size and modification time; identical files at different paths are not merged. Deleted records, missing files or changed content may be saved again. Deduplication applies within the current data directory. Changing methods does not rescan old images or clean up existing duplicate records.

Why does an image saved with the paste shortcut have no OCR text?

Images go to Pasted content by default. To send new images to Image tools, enable Save images to Image Processing in Content collection → Pasted content. Automatic recognition also depends on Enable OCR, Auto-OCR pasted images and an available OCR engine. Existing images are not moved, and duplicates are not recognized again just because you change the destination. Images collected automatically in the background still do not run OCR automatically.

Can I import several images at once?

Select or drop multiple images into Image tools, save them first, then run batch OCR when needed. Distinct records saved successfully in the batch can form a group automatically. Existing group members keep their original membership. PNG, JPEG, BMP and the first image of TIFF are supported. A batch allows up to 100 images, with a 96 MiB per-image and 512 MiB total limit, also subject to image dimensions and decoding memory. Importing does not run OCR. Reimports reuse valid records and existing recognition results. Stopping keeps saved images; retry failed or unfinished items and add successful records to the same batch group.

How do I collect several items and copy them together?

Enable the Content recording shortcut in Settings → Screenshot and choose the sources to include. Press it once to start, collect content as usual, then press it again to finish and copy the combined output. Only screenshots are selected by default. Choose other sources separately; automatic sources also require auto-collection to be enabled. Text keeps its body, while images and file references from other sources provide local paths. Screenshots use the selected output, with OCR required for text output. This does not stitch images together. Apps receiving paths must be able to read local files. Clearing the current list does not delete saved history.

How do I capture a fleeting moment on screen?

Enable Freeze screen when capturing in Settings → Screenshot → Behavior. Triggering capture takes a snapshot before you select, save or annotate a region. The original app or video keeps running. This option is off by default. Select within any one screen; a single selection cannot span multiple screens. Start a new capture if the monitor layout changes. Scrolling capture still uses live frames while scrolling.

How do I extract text from an image? Do I need a model?

No additional model is needed. Add an image in Image tools and use built-in OCR to extract text. For a PDF, open PDF recognition. Recognition runs on your computer. Built-in OCR extracts plain text. For tables or formulas, set up an external model that accepts images. Batch import and background auto-collection save images without starting recognition automatically. New unrecognized images received from devices into Image tools have a separate automatic OCR setting, on by default. OCR must also be enabled and its current engine available. Turn this off in Settings → OCR → Auto-OCR received images. Remote engines send the image according to their configuration.

Can I correct OCR text and keep the original result?

Yes. Choose Revise text in the recognition results, edit the selected version and save a new revision. The original result, image and existing organization are preserved, so you can switch versions to compare them. When editing Markdown, keep any formatting markers you still need.

How do I recognize tables and formulas?
How do I give a screenshot or annotated image to AI?

After capturing, open the annotation editor, mark the relevant area and add your instructions. Choose Copy annotated image path from the copy menu. The app saves the annotated image and editable annotations before copying its path. Paste the path into an AI app that can read local images, add your question and submit it. The path is not a command to run, and SanFrame does not send it for you. You can also copy the image itself if the AI app accepts pasted images. The original image and current annotations are kept, but this does not create a separate historical version for every annotation edit.

Saving and organizing

Can I delete the original after saving a file reference?

Saving a reference is not a backup. File references primarily store file locations. To retain a copy, select Backup copy in file-reference settings and confirm the copy was created. Add several references at once to automatically group distinct records saved successfully. Dragging files requires a readable original path. If the path cannot be obtained, follow the prompt to copy files from your file manager, then save their references. Existing group members keep their original membership. Moving or deleting the original can break the reference. When the external original is missing, choose a new location for the existing reference while retaining tags, favorites, groups and backups. This updates the reference without moving files or replacing an existing backup.

How do I export my material? Can I restore it into SanFrame?

In Settings → Data → Export history, choose content types, record scope and dates, then select an export format. Text export saves text and file paths, without original images or attachments. ZIP export includes readable originals, edited images, pasted-content files, referenced files and backups within the selected scope. Files that could not be read are listed inside the archive. Exports do not contain the full recognition-version history, app settings or group relationships, and cannot be imported directly to restore the app. They are not a complete app backup. Pasted text exports contain current content only, without saved versions. Use a complete data-directory backup to preserve versions, the current selection and settings.

Where is my data stored? Can I move it to another drive?

View the current directory in Settings → Data → Storage. It holds settings, history and files saved by SanFrame, such as images. External file references still point to their original locations. Use Change to move the data. Saving pauses while the app copies and verifies it before switching. The original directory is kept, and any existing SanFrame data at the destination is preserved separately rather than merged. The app reloads when the move finishes.

What can I do with tags, and how do I organize them?

Use ordinary tags for your own categories, such as Reading or Project material. A record can have several tags; add them manually to images, pasted content and file references. Automatically generated app-name tags and tags assigned to apps are ordinary tags too. Click a record tag to filter related material, or open a workspace from its count in Manage tags, including archived records. Multiple tags match any selected tag, alongside keywords and favorites. Searching image text needs saved OCR; filtering by tag does not. Auto-collected, Synced, Transferred, Note and Private are system tags for sources, states or protection. An ordinary tag with the same name does not gain those properties. In Display order, arrange ordinary and system tags together; overflow appears under +N. Hiding a tag does not change its records or filtering. Tags also work as sending conditions and AI scope filters. Sharing tags and opening records remain separate settings; an ordinary tag does not automatically grant AI access.

Can I edit pasted text and keep earlier versions?

Yes. Open pasted text in history, favorites or archives and edit it. Saving keeps the same record, original collection time, tags, groups and favorite state. Enable Save edit versions in Settings → Content collection → Pasted content. It starts off. The first changed save preserves the original, and later changed saves add versions. Selecting a version makes it current, including after reopening or restarting. Editing an older version adds a new version. Turning saving off keeps existing versions. If current content has no saved snapshot, switching warns that it cannot be restored from the version list; preserve anything you need first. Only non-current versions can be deleted individually. Deleting a record also deletes all its saved versions. Copy, search, ordinary export, device transfer, sync and AI reads use current content only. Version history stays local; ordinary exports are not full recovery backups. AI edits use separate operation history without adding manual versions. Switching versions or revoking access cannot recall text already sent.

Can I add notes, and are they shared with devices or AI?

Yes. Use a record’s toolbar or context menu to add, view or edit a multiline note on an image, pasted item or file reference. Notes are stored separately without changing the original text. The same record shares its note across history, favorites, archives and groups, and gains the Note system tag. Clicking the tag filters records; use the note action to read it. Notes stay local by default. Enable note sync separately for each content category to share them with permitted records. Add notes under sending conditions only selects the record; it does not grant note-text sharing. Manual device transfers and ordinary exports omit notes. AI note reading needs independent permission, respects record scope and is read-only. Notes on private records or records awaiting privacy checks are always withheld, even when redacted body text is allowed. Disabling sharing or access cannot recall content already sent.

How do I keep frequently used material easy to reach?

Add it to favorites or pin it; pinning does not require a favorite first. History, favorites and archives manage their pins separately, and you can pin a whole group. Only records pinned in ordinary history, and their groups, skip automatic archiving. Pins in favorites or archives affect the order in that view only. Select several records to apply favorites, pins or tags together, with actions offered for the selected content types. Both single-record and bulk toolbars let you arrange frequently used icons and their order.

How do I keep related material together?

Images, pasted text and file references can share one group. Keep a task’s captures, explanation and attachment locations together, or group text as Today’s to-dos and pin the next item until handled. Original records remain; file references still depend on the originals or saved backups. Choose Create group in the list header, continue selecting across types, then review and finish. You can also add to an existing group. Switching types, filters or browsing modes preserves the current group draft. Names, membership and order update together; favoriting, pinning or archiving a member does not split the group. Unified list combines the three types and shows a group once. By type keeps separate workspaces, with each relevant type opening the same full group. Viewing all members does not select hidden members, restore archives or expand AI permissions. Transfer entire group for a group of one type sends all members, including those hidden by the current filter, and retains the name and order on the receiving computer. Sending only selected members creates independent records on the receiver. Mixed-type groups also transfer selected content only, without group relationships. Mixed-type relationships stay local and cannot yet be changed directly by AI. Whole groups transfer between computers only. Received copies are independent; later edits do not write back to the sender.

Will older history be deleted automatically?

Exceeding a retention limit does not automatically delete records. Images, pasted content and file references can each be archived by count or age and remain accessible afterward. Records pinned in ordinary history, and their groups, skip automatic archiving. Manually deleting a record affects this computer only and cannot be undone. It does not recall copies already received by other devices. Whether associated files are also cleaned up depends on the deletion settings and confirmation shown. Archive or restore selected members without breaking their groups. Restored records return to history by restoration time, with original creation dates retained in details. Opening a group does not restore archived members. Restoration still follows each type’s retention limits; follow the actual notice if capacity is insufficient.

How do I remove unwanted auto-collected records?

Filter by Auto-collected, select the records in the list and use Delete. Check the selection first; deleting records cannot be undone. Images, backups and received copies saved by SanFrame follow the relevant deletion settings. Referenced external originals are kept. To inspect leftover internal files, open Settings → Data → File cleanup. If you authorize AI to suggest cleanup, it only marks candidates. You still decide what to delete.

Can I continue working on a PDF later?

Yes. PDF projects are kept when a task finishes, pauses or closes, so you can reopen them to continue reading or recognition. Select pages in the page list to recognize, proofread or export them. Batch recognition skips completed pages; individual pages can be recognized again while retaining versions. Original / Processed lets you compare preprocessing without starting recognition or changing the original PDF. To remove a project, confirm its deletion in Manage PDF projects. Its recognized text and proofreading versions are removed and cannot be restored. The original PDF and exported files remain, and deletion may not free all occupied space immediately.

Device connections and transfer

How do I connect computers or an Android phone?

Open SanFrame on each computer and SanFrame Transfer on the Android phone, all on the same local network. Enable local-network access, invite a device from the computer’s device list, compare the pairing codes and confirm on each device. Once connected, choose recipients in Device transfer and send your content. Files copied in Explorer or Finder can also be sent with Paste and send. A path copied as ordinary text is still treated as text. If a device is missing, check that both apps are running, local-network access is enabled and network permission is allowed, then search again or use an invitation link. If you see a version warning, update all participating devices, including the one forwarding content, and reconnect. You do not need to delete history.

Will paired devices reconnect automatically next time?

Yes. After the first pairing, computers running SanFrame on the same local network can restore their previous connection. Connect automatically on startup and Reconnect after an unexpected disconnection are enabled by default. To reconnect after signing in to your computer, also enable Launch at login on each computer. That setting is off by default. Choosing Disconnect pauses automatic reconnection, even after restarting the app. Select Rejoin group to resume. Closing the app normally, shutting down or an unexpected network loss does not remove reconnection eligibility. Leave sync group requires a new invitation. Received content is kept.

Can I sync only tagged material, and how do I send older records?

Yes. In sync settings, choose Send after marking under Sending from this device, enable Add tags, and choose all custom tags or specific tags. During an active connection, adding a selected tag yourself or an app rule tagging newly collected content can trigger sending. You can also choose favorites, pins, grouping or adding a note; matching any selected condition is enough. The default sending mode remains sending new content. The shared content scope is shared by the group; changes by any member affect everyone. Sending from this device controls only this computer. Images and their OCR, favorites and saved edits start enabled. Pasted content, file references, auto-collected sources, groups, ordinary tags, pin states and notes require separate choices. Selecting a record by its tag or note does not itself share the tag or note text; shared content controls those separately. Private tags, archive locations and manual pin order stay local. Sending after marking normally selects only records added during this connection. For older material, enable Allow content from before connecting, then explicitly apply a selected mark. Enabling options, changing rules, reconnecting or adding a device does not backfill history or offline content. App rules only trigger new content; existing tags, duplicate collection and received writes do not backfill. Selected tags can send new auto-collected records without enabling the entire auto-collected category. Joining a group normally selects only newly added members; optionally include other members, subject to category and older-content rules. Removing the last matching condition stops further updates while keeping received copies. You can also send existing material manually; manually sent copies do not join ongoing automatic sync. File references send file copies without tracking later edits to the original. The limit is 8 GiB per file, and folders cannot be sent directly. Syncing existing OCR results is separate from automatic OCR on receipt. The receiving computer may recognize newly received unrecognized images, subject to its own OCR switch, receiving option and engine settings; existing results are skipped. Pasted text syncs current content only, without local saved versions. Mixed-type group relationships currently stay local. Sending members still follows their content types and sending rules; do not expect a complete mixed-type group on another computer.

Will incoming text replace what is already there?

Yes. Incoming text replaces the entire current text in Device transfer. Clearing text is shared too; edits from different devices are not merged. Live sending is on by default. Editing text or actively changing the recipient selection sends the current text to those devices. To finish writing before sending, turn live sending off and send manually. Receiving text does not automatically send it back. History saving is separate: when enabled, newly received nonblank content changes are saved independently. Later text or clearing the editor does not overwrite earlier history. Consecutive updates differing only in leading or trailing whitespace reuse the previous valid record; this is not a keystroke archive.

Where are received text, images and files saved?

Supported static PNG, JPEG, BMP and WebP images enter Image tools; other files enter File references. Animated and unsupported images cannot be transferred as still images. Settings → OCR → Auto-OCR received images starts on, and also requires OCR to be enabled with an available engine. New unrecognized images queue in order. Images with existing text, versions or a no-text result are skipped; enabling the option does not scan old history. You can disable or stop it and retry failures manually, keeping the original. Online engines send images to the configured service. Live text stays in Device transfer by default. Enable Save received text to keep subsequent nonblank content changes as separate Pasted content records, preserving earlier text, tags and favorites. Rapid typing may be combined before delivery, so this is not a keystroke history. Consecutive updates from the same device in one connection reuse the previous valid record when only leading or trailing whitespace, BOM or zero-width spaces differ. Internal whitespace, case and punctuation remain significant. Clearing or blank text neither creates an empty record nor erases history. Turning saving off keeps history; re-enabling does not backfill. If the old record was deleted, archived or edited locally, later new messages are saved separately.

How can I paste received content straight into another app?

Enable the option for the output you need: Text itself: turn on Copy received text in Device transfer. It preserves line breaks and does not require saving text history. Blank text does not clear your clipboard. Paths for manually sent images or files: turn on Copy received paths. This also covers material sent manually from history, favorites, archives or groups. A pasted-text record provides the path of its saved .txt file. Paths for automatically synced material: configure Image tools, Pasted content and File references separately in Local network sync → Sync & receive → Receiving. These options are off by default and affect future arrivals only. Copied paths belong to the receiving computer; they are not sender paths or image pixels.

AI and privacy

What can a connected AI help me do?

Both the built-in assistant and external MCP can find records, read saved text, add content, organize tags, groups, favorites and pins, and search app guides within your permissions. Editing text, removing tags or unfavoriting needs separate authorization. You initiate tasks; SanFrame does not run continuous background organization. Full-text reading uses saved content without starting OCR. AI cannot read PDF projects or file-reference bodies, permanently delete records, edit external source files or expand its permissions. Pasted-text version history is not available to AI. Images need a model that supports them and permission for the corresponding workspace. The built-in assistant sends permitted images to DeepSeek as needed; external MCP returns local image paths. Allowing or selecting a record does not itself upload it. With transfer permission and recipients connected beforehand, AI can send text and files. Filling text follows the live-send setting. A submitted request does not prove saving or reading by the recipient, and received copies cannot be recalled. AI can organize supported groups of one type. New mixed-type group relationships require manual management; member content still follows its own permissions.

How do I use the built-in assistant or connect external AI?

Open AI access from the top toolbar. The built-in assistant and external AI start off and have independent switches. They share Data scope, Action permissions and Operation history. Initially only Device transfer is selected; authorize workspaces, full text, images, notes and actions as needed. Built-in assistant: open Model settings, enter your own DeepSeek API key and choose Save and enable for first setup, then ask a question in AI Assistant. External MCP need not be enabled. The interface currently uses DeepSeek’s official service and does not offer arbitrary service addresses. External AI: enable External AI access at the bottom and copy the MCP configuration into an app that supports MCP and can reach this computer’s local address. The configuration includes a connection key; share it only with trusted software and keep SanFrame running. Turning off either route leaves shared permissions and the other route unchanged. Disabling the built-in assistant stops its current response while keeping models, keys, conversations and operation history. Clearing the key also disables it; sent content cannot be recalled. A fixed tool list is not blanket authorization. Every call checks scope. Ordinary permission changes usually need no reconnect; refresh or reconnect after upgrades if tools are cached. Update the configuration when the port or access token changes. Archive conversations from the sidebar and restore them from Archived later, keeping messages and drafts. Messages show times and can be copied or reused as a new question without rewriting the original. Conversations and archives stay local, do not sync to devices and do not change record permissions.

Which AI app or model should I choose for organization?
Can I ask how SanFrame works without sharing records?

Yes. Choose Guide only in the built-in assistant’s input area. That conversation uses app guides and permitted settings diagnosis, without record-reading, organizing or transfer tools. It does not change global permissions. The model still needs internet access and your question goes to DeepSeek, so do not type material you want to keep private. Settings diagnosis is allowed by default and shared by the built-in assistant and external MCP. You can turn it off in AI access from the top toolbar. It returns switches, modes and counts, without paths, keys, device names, tag names or record bodies. Answers can link to relevant settings; you change the settings and permissions yourself. The assistant cannot change them automatically. External MCP can also search guides, but Guide only restricts the built-in conversation, not other actions already authorized for the external app.

How do I limit what AI can see?

The built-in assistant and external MCP share the same record scope. In AI access, choose workspaces, record dates, custom tags, favorites, groups, archives and sources. File references can also be limited by folder. Any matching exclusion blocks a record; another selection cannot add it back. For example, excluding received records also excludes received records you have favorited. Records without custom tags are not restricted by the custom-tag selection alone. Turning off full-text reading may still allow short summaries in authorized record queries. To provide none of a record’s content, exclude it from the scope or add the system Private tag. Date filters save fixed dates; choosing Today or Last 7 days does not create a continuously moving range. These settings govern access through SanFrame only. They cannot restrict the AI app’s own file or terminal tools. File-reference folder filters also do not restrict ordinary files that are not associated with SanFrame records. The built-in assistant sends images within scope to the model as needed; external MCP returns local paths. Changing shared data permissions or record scope stops the current built-in response, but cannot recall requests already sent.

How do I keep private material from AI? Can I share redacted text only?

Add the system Private tag to a record to block AI access by default. Creating an ordinary tag with the same name does not give it this protection. You can instead choose Redacted text only in Record scope → Privacy and authorize text reading for the relevant workspace. Sensitive passages are replaced only when they can be located reliably; otherwise, the record stays blocked. Original images, paths, older versions, notes and organization or editing of that private record remain unavailable. Local Blur all content / Blur sensitive text settings control display separately from AI permissions. They do not encrypt files or redact image pixels. Automatic checks can miss sensitive content, and redaction does not guarantee anonymity.

Can SanFrame tag private content automatically?

Yes. In Privacy management, enable automatic tagging separately for screenshots/images and copied/pasted content. Both are off by default. Choose basic rules, the local AstrLink Guard model or Rules + model. Guard is about 39 MB and must be explicitly downloaded and enabled. No separate Python installation or model service is needed. Checks run locally without uploading records, but they depend on saved text; images without OCR are not automatically recognized. Choose model detection types such as names, addresses, contact details, accounts and passwords. Changes affect future checks without rescanning history or removing existing tags. Start a separate batch check for older material, and use recent markings to view recorded reasons. Rules and models can misclassify or miss content. Remove mistaken tags in the original workspace. Downloading the model does not enable collection or scan history automatically.

Can I undo AI changes I do not like?

Changes from the built-in assistant and external AI share operation history, with their source identified. Authorized changes run immediately. Review them afterward in AI access → Operation history, then confirm that you want to keep them or undo them. Eligible changes can still be undone after confirmation. Later edits, dependencies from other operations or removal of recovery data may make undo unavailable. Clearing operation history does not delete your material, but ends undo for the affected changes. Undoing local changes does not stop later tasks in the external AI app or recall content already provided to AI or sent to another device.

Troubleshooting

Which permissions does Mac screen capture need?

Allow SanFrame screen-recording access when macOS prompts you. You can also enable it in System Settings → Privacy & Security → Screen Recording or Screen & System Audio Recording. Return to the app and check again; reopen SanFrame if the permission has not taken effect. Automatic scrolling capture also needs Accessibility permission. Grant access to the SanFrame app you are currently running.

How do I resume a paused scrolling capture?

A pause does not necessarily mean the page has reached its end. Follow the prompt to scroll the page or drag the selection’s edge up or down until it overlaps captured content, then choose Continue. Moving the selection preserves its size and horizontal position and stays within the current screen. Without reliable overlap, the app keeps the existing capture instead of forcing a join. At the bottom of a page, you can move the selection down to include remaining visible content, then choose Finish.

How do I troubleshoot a problem?

Start with the error shown for the current action. If a shortcut cannot be enabled, change the combination or check it again in its settings row. If auto-collection saves nothing, check that the app is running, collection and the relevant content types are enabled, and no read or save error is shown. If the problem remains, view or export logs in Settings → Maintenance → Diagnostics. App error logging is on by default. Capture logging is off by default, so enable it before reproducing a capture problem. Logs stay local and do not record images, OCR bodies, clipboard content or user file paths. Capacity and retention limits apply. When reporting a problem, include the app version, operating system, steps to reproduce it and relevant logs. An empty log does not prove that no failure occurred.