Home
TR
Reference

Interface Reference

Every control in the Colpan window — what each button, field and panel does.

Chat (AI)

  • Model picker — Switch the AI model for this conversation. The list shows every configured cloud and local provider plus any embedded model. The picker locks once a chat has started.
  • Select model — Pick this provider and model for the chat. The dot marks the active choice; selecting closes the list.
  • Review attachment — Reopen the preview for this attachment so you can re-check its contents and PII before sending.
  • Remove attachment — Drop this attachment from the message and delete its temporary copy. It will not be sent.
  • Cancel — Close this dialog without changing the PII term list.
  • Remove term — Drop this term from the PII list so future scans stop flagging it. Text already masked in past messages is unaffected.
  • Add term — Save this term as PII so every future scan flags and masks it before text leaves the device.
  • Open conversation — Load this past conversation into the chat. Your current chat is saved first, then replaced.
  • Cancel send — Stop sending and return to your draft so you can edit it. Nothing is sent.
  • Send raw — Send the message unchanged, including the detected personal data. Use only when the text is safe to share as-is.
  • Send anonymized — Replace each detected item with a vault token, then send. The model never sees the original personal data.
  • Keep or mask — Toggle this detection. Strike it through to send the original word, or restore it to mask it. Use this to drop false positives.
  • Fix the error — Apply the suggested fix for the error above, such as adding an API key or switching to a local model.
  • Dismiss error — Hide this error banner without taking any action.
  • Detected personal data — This span was flagged as personal data and will be masked before send. Click it to inspect the match or remove the flag.
  • Remove from PII — Stop treating this span as personal data. It will be sent as written and no longer masked.
  • Close — Close this popup without changing the flag on the span.
  • Chat history — Open a list of your past conversations. Pick one to load it back into the chat.
  • PII shield — Turn automatic masking on or off for this cloud chat. When on, detected personal data is replaced with vault tokens before the message is sent.
  • New chat — Save the current conversation and start a fresh one. Disabled while the chat is already empty.
  • Send message — Send your message to the AI (or press Enter). While a reply is streaming this button stops the wait instead.

Cloud Sources

  • Close — Close the Cloud Sources window. Saved sources stay configured; unsaved edits in the form are discarded.
  • Cloud source — Click to load this cloud source into the form on the right so you can review or edit its settings.
  • New source — Clear the form and detach the selection so the next Save creates a brand-new cloud source instead of editing an existing one.
  • Scheme — Pick the cloud backend (Google Drive, S3, WebDAV, OneDrive, Dropbox). Changing it swaps the option fields below to match that provider.
  • Fetch mode — Choose how files are pulled: persistent_cache keeps a local copy on disk, transient_fetch downloads on demand and does not retain it.
  • Credentials guide — Open the step-by-step guide for obtaining the API keys or tokens this cloud scheme needs, in your browser.
  • Sign in — Run the OAuth browser sign-in for this source (Google Drive, OneDrive, Dropbox). A browser tab opens; approve access and the token is saved here.
  • Test connection — Try connecting to this cloud source with the current credentials. The result shows on the left without saving anything.
  • Delete source — Remove the selected cloud source from config.toml. The indexed data stays; only the connection is detached.
  • Save source — Write the form to config.toml, updating the selected source, or creating a new one if none is selected.
  • Browse models — Open the OpenRouter model catalog in your browser. Sign in there, find a model, then paste its slug into the field above.
  • Account and API keys — Open your OpenRouter account and API-keys page in the browser to create or copy the key this picker uses.
  • Active profile — Switch which saved BYOK profile is active. The edit form below updates to show the selected profile’s key, service and models.
  • New profile — Start a blank draft profile. Nothing is saved until you click Save; Cancel discards it and brings back the previous profile.
  • Clone profile — Duplicate the active profile, including its key and models, as a new profile you can rename and tweak independently.
  • Delete profile — Remove the active profile permanently. Disabled when only one profile is left, since at least one must remain.
  • Service — Choose the provider this profile talks to (OpenRouter, OpenAI, Anthropic, Google, or a custom OpenAI-compatible endpoint). Changing it updates the key placeholder and shortcut links.
  • Active model — Make this saved model the active one for the profile. The active model is what chat requests use until you pick another.
  • Rename model — Load this model’s slug and token cap into the edit row below so you can change them in place, then click Update.
  • Delete model — Remove this model from the profile’s saved list. If it was the active model, the profile falls back to the next saved entry.
  • Add or update model — Save the slug in the field with its token cap onto this profile, adding a new model, or replacing the one you are renaming.
  • Cancel rename — Stop renaming and clear the edit row without changing the saved model list.
  • Token preset — Fill the token-cap box with this common context size in one click, so the next Add or Update uses it.
  • Open your OpenRouter account and API-keys page in the browser to create or copy the key this profile uses.
  • Browse OpenAI models — Open the OpenAI models documentation in your browser to find a model name, then type it into the model field.
  • Your API keys — Open your OpenAI API-keys page in the browser to create or copy the key this profile uses.
  • Browse Anthropic models — Open OpenRouter’s Anthropic-vendor catalog in your browser, then copy a slug like ‘anthropic/claude-…’ into the model field.
  • OpenRouter API keys — Open the OpenRouter API-keys page in your browser. Even on this vendor tag, requests route through OpenRouter today, so the key lives here.
  • Browse Google models — Open OpenRouter’s Google-vendor catalog in your browser, then copy a slug like ‘google/gemini-…’ into the model field.
  • Discard draft — Throw away this new-profile draft without saving anything and return to the previously active profile.
  • Save profile — Commit this draft to your saved profiles and make it active. Stays disabled until the required fields (API key, and Base URL for custom) are filled.
  • Close the downloads window. Any installs in progress keep running in the background.
  • Close the AI chat window. Your prompt and the suggestions are discarded.
  • Ask — Send your plain-language prompt to the assistant. It returns up to five CQL queries you can click to run.
  • Use this suggestion — Click to drop this CQL query into the search bar and run it. The pill on the right shows how many results it would return.
  • Clear all history — Erase every recent search from this list. Saved (pinned) searches are kept; only the unpinned history is removed.
  • Close the search-history dropdown without running anything.
  • All searches — Show every recent query you have run, newest first. This is the default view.
  • Saved searches — Show only the searches you pinned with the star. These stay until you unpin them.
  • Run this search — Click the row to load this past query into the search bar and run it again.
  • Pin search — Toggle this query in your Saved searches. Pinned queries survive a history clear and show under the Saved tab.
  • Remove search — Delete this single entry. In the Saved view it also unpins it and removes every matching history row across profiles.
  • Cancel — Discard the staged document and close this preview without sending anything to the chat.
  • Send to chat — Send this document to the chat with the PII shield applied. If the shield is off, you confirm once more before the original text goes through.
  • Close the Qdrant experiment window. The service and any transfer keep running in the background.
  • Start service — Launch the local Qdrant vector database so you can transfer profiles into it and run semantic queries.
  • Stop service — Shut down the local Qdrant service. Transferred collections stay on disk and come back when you start it again.
  • Refresh status — Re-check whether the Qdrant service is running and update the status line above.
  • Source profile — Pick which existing Colpan profile’s indexed content gets embedded and copied into the Qdrant collection.
  • Transfer — Embed the selected profile’s documents on the CPU and load them into Qdrant. If the collection already exists you are asked before it is overwritten.
  • Cancel transfer — Keep the existing Qdrant collection untouched and abandon this transfer. Your profile choice is preserved.
  • Overwrite collection — Delete the existing collection and rebuild it from the selected profile. This cannot be undone.

Dialogs

  • Cancel — Close this dialog without making any changes.
  • Confirm — Accept the name you typed and continue.
  • Delete profile — Permanently remove this profile and every index inside it. This cannot be undone.
  • Create profile — Create the new profile with the name you typed and switch to it.
  • Rebuild from scratch — Wipe the current index and reindex every file from the start, instead of only the new or changed ones.
  • Keep deleted files — Leave entries for files that no longer exist on disk in the index, instead of removing them during this run.
  • Edit folders — Review and change the folders this index scans before indexing starts.
  • Edit filters — Open the filter settings for this index to choose which file types and extensions get indexed.
  • Close this dialog without starting indexing.
  • Start indexing — Begin scanning the folders and extracting text from documents with the options chosen above.
  • Skip file — Stop working on the file in progress and move on. The skipped file is not added to the index.
  • Stop indexing — Halt the current indexing run. Files already indexed stay; the rest are left for next time.
  • Hide window — Close this window. Indexing keeps running in the background and you can reopen it from the status bar.
  • Keep file — Do not skip. Close this prompt and keep indexing the current file.
  • Confirm skip — Skip the current file for good. It will not be added to the index, and indexing continues with the next file.
  • Local directory — Pick a folder on this computer to add to the index. Opens your system folder browser.
  • Cloud source — Choose which configured cloud account (Drive, OneDrive, S3, WebDAV) the remote folder belongs to.
  • Add cloud directory — Add the remote folder at the path you typed from the selected cloud source to this index.
  • Close this dialog without adding a directory.
  • Create — Create the new collection (or category) with the name you typed.
  • Rename — Save the new name for this collection (or category), replacing the old one.
  • Archive scope — Choose which archive formats are opened and indexed for this file type. Global follows the index-wide limit; None skips archives entirely.
  • Global archive scope — Set the default for every file type: which archive formats are opened and indexed unless a specific type overrides it.
  • Pick a day — Click a day to set the start of the range, then click another for the end. The two clicks mark the From and To dates.
  • Previous month — Show the month before the one displayed. Your selected dates are kept.
  • Next month — Show the month after the one displayed. Your selected dates are kept.

Filters & PII

  • Back — Leave the filter presets editor and return to the previous panel.
  • Filter preset — Select this preset to edit its extensions and size limits below.
  • Add preset — Create a new custom filter preset you can rename and fill with your own extensions.
  • Remove extension — Drop this file extension from the selected preset.
  • Add extension — Add the typed extension to the selected preset. Files with that extension will then be indexed.
  • Active preset — Pick which filter preset you are editing. The extension list and size limits below switch to match it.
  • Toggle extension — Turn indexing of this file extension on or off for the active filter. Built-in filters are read-only.
  • Add custom extension — Add the typed extension to this filter so matching files get indexed.
  • Close — Close this window. Extension and size changes are saved as you make them.
  • Close the filter editor. Your extension and limit changes are saved automatically.
  • Index — Choose which index this filter applies to. The extension list below belongs to the selected index.
  • Apply preset — Replace this filter’s entire extension list with the chosen preset. You will be asked to confirm first.
  • Hide unused — Collapse parsers that have no extensions enabled, so the catalog only shows what you actually use.
  • Clear search — Empty the catalog search box and show every parser and extension again.
  • Parser group — Expand or collapse this parser to show or hide the file extensions it handles.
  • Add or remove this extension from the filter on the right. Checked extensions get indexed by this parser.
  • Add new extension — Open a dialog to register a brand new file extension and assign it a parser.
  • Copy and edit — Make an editable copy of this read-only built-in filter so you can change its extensions and limits.
  • Clear overrides — Reset every per-extension override (parser, archive, OCR, size) back to the filter’s global defaults. The extensions themselves stay.
  • Remove all extensions — Empty this filter’s whole extension list. You will be asked to confirm; the index must be rebuilt afterwards.
  • Parser — Pick which backend extracts text from this extension. Only parsers that support the format are listed.
  • Archive scope — Choose which archive types are opened and indexed for this extension. Global follows the filter default below.
  • OCR engine — Pick the text-recognition engine for scanned content in this extension. Locked to a dash when the format cannot be OCR’d.
  • Minimum size — Skip files of this extension that are smaller than the chosen size. A dash means no minimum.
  • Maximum size — Skip files of this extension that are larger than the chosen size. A dash means no maximum.
  • Global minimum size — Default smallest file size to index, applied to every extension that has no per-extension minimum of its own.
  • Global maximum size — Default largest file size to index, applied to every extension that has no per-extension maximum of its own.
  • Extraction timeout — How long the extractor may spend on a single file before it is given up on and skipped.
  • Global archive scope — Default set of archive types opened and indexed across this filter, unless an extension overrides it in the table above.
  • Take this extension out of the filter. Its files will no longer be indexed.
  • Cancel — Dismiss this dialog without changing anything.
  • Confirm replace — Apply the destructive change. The extension list is overwritten and the index must be rebuilt; this cannot be undone.
  • Pick the backend that will extract text from the new extension.
  • Close this dialog without adding the extension.
  • Register the typed extension with the chosen parser. You can then enable it for filters from the catalog.
  • Select entry — Tick this entry to include it in a bulk action such as Delete at the bottom of the vault.
  • Edit entry — Change the original value this token maps to. The masked token itself stays the same.
  • Delete entry — Remove this entry from the vault. The token will no longer un-mask back to its original value.
  • Save changes — Store the edited original value for this entry.
  • Cancel edit — Discard your edits and keep the original value unchanged.
  • Close the PII vault. Your entries stay saved.
  • Add entry — Show or hide a small form for adding a new masking entry by hand.
  • Import CSV — Load masking entries from a CSV file and merge them into the vault.
  • Export CSV — Save every vault entry to a CSV file so you can back it up or move it to another machine.
  • Save new entry — Add the typed value under the chosen category as a new masking entry in the vault.
  • Show all categories — List vault entries from every category at once.
  • Filter by category — Show only the vault entries that belong to this category.
  • Select page — Tick every entry shown on the current page so you can act on them together.
  • Clear selection — Untick every selected entry across all pages.
  • Delete selected — Permanently remove every ticked entry from the vault at once.
  • Previous page — Show the previous page of vault entries.
  • Next page — Show the next page of vault entries.
  • Cancel scan — Stop the running PII scan. Any matches found so far are discarded.

Index Panel

  • Collapse panel — Fold the index list away to free up space, or expand it again. The panel keeps a compact summary while collapsed.
  • Indexing progress — How far the current indexing run has got. Click to reopen the detailed indexing window if you closed it.
  • OCR backfill — Text from scanned images is still being extracted in the background. The number is how many documents are left. Click to open the indexing window.
  • Stop indexing — Halt the indexing run in progress. Documents already indexed are kept; the rest are left for the next run.
  • Add index — Create a new index. You pick the folders to scan, and Colpan builds a searchable catalog of the documents inside.
  • Panel menu — More actions for the whole panel: add an index, index everything, manage indexes, or open the overview.
  • Add first index — Get started by creating your first index. Pick the folders to scan and Colpan builds a searchable catalog of the documents inside.
  • Index row — One index and its document count. Click to expand its folders, or right-click to open the index directory.
  • Enable index — Include this index in searches. When unchecked the index is skipped and its documents will not appear in results.
  • File-type filter — The file types this index keeps. Click to see the full extension list and edit the filter.
  • Index actions — Actions for this index: add or exclude a directory, manage it, reindex, or delete it.
  • Include directory — Scan this folder as part of the index. When unchecked the folder is skipped and its files are not indexed.

Indexing

  • Stop indexing — Halt the running indexing job. Files already indexed are kept; the rest are left unprocessed.
  • Close panel — Hide this indexing panel. Indexing keeps running in the background and you can reopen the panel later.
  • Close window — While indexing is running this only hides the window and the job continues; once it is finished this closes the window.
  • Save and close — Save all changes to this index and close the window. Edits to name, directories and options take effect immediately.
  • Format preset — Pick which file types this index covers, such as documents or images. Click to open the preset list.
  • Watch mode — Choose how this index keeps up with changes: continuous watches folders live, manual updates only on demand, disabled stops watching.
  • Remove tag — Detach this tag from the index. Tags help you group and find indexes; removing one does not touch indexed files.
  • Add directory — Pick a folder to include in this index. Its files are scanned and indexed on the next run.
  • Enable directory — Turn this folder on or off for the index. When off, its files are skipped on the next run without being removed.
  • Exclude directory — Pick a folder to leave out of this index. Files inside it are skipped even if a parent folder is included.
  • Remove exclusion — Stop excluding this folder. Its files become eligible for indexing again on the next run.
  • Filter preset — Select this set of file types for the index. The summary line shows which extensions it includes.
  • Index selector — Choose which index to inspect. The directory tree and file list reload to show that index’s contents.
  • Directory — Click to list this folder’s indexed files on the right. Double-click to expand or collapse its subfolders.
  • Expand folder — Show or hide the subfolders nested under this directory.
  • Indexed file — Double-click to open this file in its default application. The row shows the file type and name as stored in the index.
  • Close viewer — Close the database viewer. Your indexes are not changed; this only closes the window.
  • Rebuild from scratch — Wipe everything already indexed and index all files again. Slower, but clears any stale data.
  • Keep deleted entries — Leave index entries in place for files that no longer exist on disk, instead of pruning them.
  • Directories — Open the directory list for this index so you can add or remove folders before indexing.
  • Extensions — Open the extension filter so you can choose which file types this index includes before indexing.
  • Cancel — Close this dialog without starting indexing. Nothing is changed.
  • Start indexing — Begin scanning the directories and extracting text from documents with the options above. Progress opens in a separate window.
  • Add files — Pick documents to add to the conversion queue. You can also drag and drop files onto this window.
  • Clear list — Remove every file from the queue. The files on disk are not deleted.
  • Open converted file — Open the converted output in its default application.
  • Retry conversion — Try converting this file again after a failure.
  • Output format — Choose the target file format for the conversion. Mixing different source types limits the choice to PDF.
  • Save location — Choose the folder where converted files are written. Leave it as Same as source to write next to each original.
  • Convert all — Convert every queued file to the selected format and save it to the chosen location.
  • Close converter — Close the converter window. Files already converted stay where they were saved.
  • Details toggle — Expand to show your indexes and quick CQL search tips, or collapse to hide them.
  • Indexing summary — Totals for documents, indexes and directories. While indexing is running, click to reopen the detailed progress window.

Left Panel

  • Expand panel — Grow this tab into a full overlay window. Click outside or use the minimise button to shrink it back.
  • Close panel — Collapse the side panel content. The rail on the left stays so you can reopen any tab.
  • Resize handle — Drag left or right to set the side panel width. The new width is remembered between sessions.
  • Collapse chat — Bring the expanded chat overlay back into the compact side panel.
  • Collection — Click to expand or collapse this collection. Right-click to rename, delete, or remove missing files.
  • Starred document — Click to open this document in the preview. Right-click to reveal it in Explorer or remove the star.
  • Copy logs — Copy every log entry shown here to the clipboard, so you can paste them into a bug report.
  • Clear logs — Remove all entries from this view. New events keep arriving as the app runs.
  • Enable or disable — Click to turn this plugin on or off. Disabled plugins stay installed but are not loaded.
  • More actions — Open a small menu to reveal this plugin’s folder on disk or remove it entirely.
  • Install plugin — Pick a plugin folder to add it to Colpan. The plugin appears in the list below once it loads.
  • Open plugins folder — Open the folder where plugins live in your file manager, so you can add or inspect them by hand.
  • Category — Click to expand or collapse this category. Right-click to rename it or drop the category.
  • Saved search — Click to run this saved query in the search bar. Right-click to move it to a category or unpin it.

Management

  • Close — Close this window. Any edits you already applied to the indexes are kept.
  • Profile — Switch to this profile. Its indexes load into the list below; the highlighted chip is the active one.
  • New profile — Create a new, empty profile and switch to it. You can add indexes to it afterwards.
  • Profile menu — Open actions for the active profile: rename it or delete it.
  • Rename profile — Open a box to give the active profile a new name.
  • Delete profile — Delete the active profile and all of its index links. The default profile and the last remaining one cannot be deleted.
  • Clear search — Empty the index search box and show every index in this profile again.
  • Index — Select this index to load its settings, folders and exclusions into the editor on the right.
  • New index — Create a fresh, empty index in the active profile and open it for editing.
  • Attach index — Add an index that exists on disk but is not yet part of this profile. Disabled when there is nothing to attach.
  • Import index — Bring in a published .colpan package as a read-only index in this profile.
  • Publish index — Open the wizard to export the selected index as a shareable read-only snapshot. Select an index first.
  • Export index — Save a copy of this index to a folder you choose, so it can be moved or shared.
  • Delete index — Remove this index from the profile. You will be asked to confirm first.
  • Remove this imported index. The original package on disk is not touched.
  • Detach index — Remove this index from the profile but keep its files on disk. You can attach it again later.
  • File filter — Pick which file types this index keeps, such as Documents or Images. Files outside the filter are skipped.
  • Watch mode — Choose how this index stays current: continuous re-indexes on every change, manual only when you ask, disabled never.
  • Add local folder — Pick a folder on this computer to index. Its files are scanned into this index.
  • Add cloud folder — Index a folder from a connected cloud account. Disabled until you set up a cloud source in Settings.
  • Cloud source — Pick which connected cloud account the remote folder lives in.
  • Add the chosen cloud source and remote path to this index’s folder list.
  • Recursive — Toggle whether sub-folders of this path are indexed too. On includes everything underneath; off only the folder itself.
  • Watch folder — Toggle live watching for this path so new and changed files are picked up automatically.
  • Remove folder — Drop this folder from the index. Its files stay on disk but are no longer indexed.
  • Exclude folder — Pick a folder to skip. Nothing under it will be indexed, even if it sits inside an indexed path.
  • Exclude file — Pick a single file to skip so it is left out of this index.
  • Remove exclusion — Delete this exclude rule so the path becomes eligible for indexing again.
  • OK — Close this window. Your changes are already saved as you make them.
  • Cancel — Discard this cloud folder entry and close the form without adding anything.
  • Dismiss this prompt without making the change.
  • Confirm — Go ahead with the action described in this prompt.
  • Close the rename box and keep the current profile name.
  • Rename — Apply the new name to the active profile.
  • Browse — Open a folder picker to choose where the published snapshot is written.
  • Publish schedule — How often the snapshot is republished automatically. Off means it only updates when you publish by hand.
  • Close the wizard without publishing or saving the destination.
  • Publish now — Write a read-only snapshot to the destination right away and save these settings.
  • Save schedule — Store the destination and schedule without publishing right now.
  • Profile picker — Choose which profile to inspect. Switching it reloads the index list below for that profile.
  • Index picker — Choose which index in this profile to browse. The folder tree and file list reload to show its contents.
  • Close the database viewer and return to the main window.
  • Folder — Click to show this folder’s files on the right. Double-click to expand or collapse its sub-folders.
  • Indexed file — An indexed file in this folder. Double-click to open it in the preview window.
  • Cancel load — Stop enumerating the index. Browsing resumes once you pick an index again.
  • Close monitor — Close this resource popover. The supervisor keeps running and managing loaded engines in the background.

Result List

  • Filter and sort — Show or hide the toolbar for narrowing and sorting these results by index, type, size, date or name.
  • Index filter — Show only results that come from the selected index. Pick All to search across every index.
  • Type filter — Show only results with the chosen file type or extension. Pick All to include every type.
  • Size filter — Show only results whose file size falls in the chosen range. Pick All to ignore size.
  • Date filter — Show only results modified within the chosen time window. Pick All to ignore the date.
  • Sort order — Choose how to order the results, for example by relevance score, name, size or date.
  • Clear this filter — Reset this dropdown back to All, removing it from the result filter.
  • Search result — Double-click to open this file in the preview. Right-click for more actions like reveal, find similar and star.
  • Star — Add or remove this file from your starred collections. Starred files appear under the Starred tab.
  • Add to collection — Open the menu to add this file to an existing starred collection or create a new one.
  • CQL — Advanced query syntax. Click to open the guide with the operators, fields and filters you can insert into the search.
  • Search history — Your recent searches with result counts and timestamps. Click to pick one and run it again.
  • AI assistant — Describe what you want in plain language; the assistant turns it into a CQL query for you.
  • Vector search — Force semantic (vector) search for the next query. Finds results by meaning, not just matching keywords.
  • Search — Run the query (or press Enter). Searches every enabled index in the active profile at once.
  • Clear filters — Remove every active filter (extension, path, name and date) at once.
  • Extension filter — Limit results to files with this extension, e.g. pdf or docx. Click to edit.
  • Path filter — Limit results to files under this folder path. Click to edit.
  • Name filter — Limit results to files whose name matches this text. Click to edit.
  • Date filter — Limit results to files modified in a date range. Click to pick the dates.
  • Hover an item to see what it does. Click it to insert into the search.
  • Limit results to a file extension, for example pdf or docx.
  • Limit results to files under a folder path.
  • Limit results to files whose name matches.
  • Restrict results to a modified-date range.
  • Exclude documents that match the next term.
  • Match either alternative, for example cat or dog.
  • Separate independent query clauses.
  • Match an exact phrase, words in order.
  • Group terms together to control precedence.
  • Proximity group: every term must appear close together.
  • Proximity: terms must fall within N words of each other.
  • MoreLikeThis: find documents similar to the given text.
  • Search the document body text.
  • Search the file name only.
  • Search the full folder path.
  • Search the immediate parent folder name.
  • Search by file extension.
  • Search by document kind, such as pdf or image.
  • Search the document title.
  • Search bold or italic emphasized text.
  • Search text inside tables.
  • Search page or section headers.
  • Search page footers.
  • Search other document metadata.

Settings

  • Settings section — Switch the Settings panel to this section. The right side shows the controls that belong to it.
  • Close settings — Close the Settings window. Changes are saved as you make them, so nothing is lost.
  • Theme — Pick the colour theme. Paper is the light default, Dark is low-light, Cool is a muted blue-grey. Applies instantly.
  • Mode preset — Switch the chrome preset. Simple hides advanced controls; Advanced shows everything. Custom modes you define in config.toml appear here too.
  • Accent colour — Set the highlight colour used for buttons, links and selections. Click a swatch to apply it everywhere at once.
  • Density — Control how tightly rows and controls are packed. Compact fits more on screen; Spacious adds breathing room. Applies instantly.
  • UI scale — Zoom the whole interface up or down. Raise it on high-resolution screens for larger text; lower it to fit more content.
  • Minimum window width — Set the narrowest the app window can be dragged. The new floor takes effect after the next restart.
  • Manage profiles and indexes — Close Settings and open the Profile and Index Management window, where you create profiles and the indexes inside them.
  • Language — Choose the interface language. The change applies live; no restart needed.
  • Info tooltip level — Set how much help text the app shows. Off hides all tooltips, Standard shows short labels, Detailed adds a full explanation for every control.
  • Enable OCR — Turn on text recognition for scanned pages and image PDFs. When off, those files are indexed without their image text.
  • Manage OCR models — Jump to the Local Models tab filtered to OCR packs, where you download or remove recognition models.
  • OCR engine — Choose which recognition engine runs. Colpan OCR is built in; Tesseract and System use external engines if they are installed.
  • Save language — Save the OCR language code you typed. The new language is used on the next indexing run.
  • Force OCR on all pages — Run OCR on every page image, even ones that look text-free. Slower, but catches text the smart-skip gate would miss. Off keeps indexing fast.
  • Model preset — Pick a ready-made Colpan OCR model bundle for a script family. Choose custom to point at your own detection, recognition and dictionary files.
  • Compute device — Choose where Colpan OCR runs. Auto picks the best available; a GPU option (CUDA, DirectML, CoreML) is much faster when supported.
  • Save path — Save the file path you typed for this Colpan OCR model. It is used the next time OCR runs.
  • Correct orientation — Auto-rotate tilted or sideways scans before reading them. Improves accuracy on skewed pages at the cost of a little speed.
  • Save fallback language — Save the fallback language code used by the System OCR engine when no page language is detected.
  • Save models folder — Save the folder where OCR model packs are stored. Leave the field empty to use the platform default location.
  • Stemming — Match words by their root form, so a search for one word also finds its inflections. Turn off to match the exact word forms only.
  • Reranking — Run a second pass over the top hits so highly relevant matches can climb the list. Slightly more CPU per query.
  • Document language — Set the language whose stemmer is applied while indexing. Auto detects it from the document. Takes effect on the next indexing run.
  • Auto-scan on preview open — Scan each file for personal data the moment you open it in the preview. When off, you start the scan manually.
  • Regex detection — The built-in pattern detector that finds structured data like emails, phone numbers and ID numbers. Always on and cannot be turned off.
  • LLM detection — Use a language model to catch names, addresses and organisations the patterns miss. Needs an embedded LLM model loaded.
  • NER detection — Use a named-entity model to recognise people, places and organisations across languages. Pick a NER model in the Local Models tab first.
  • Cloud anonymize setting — Open the Chat AI tab, where the toggle for masking personal data before it is sent to a cloud model now lives.
  • Install LLM model — Jump to the Local Models tab filtered to LLM packs so you can download a model the AI (LLM) detector needs.
  • Install NER model — Jump to the Local Models tab filtered to NER packs so you can download a model the AI (NER) detector needs.
  • Manage vault — Open the encrypted PII vault, where you can review, edit or delete every stored mapping between a masked token and its original value.
  • Add to list — Add the value you typed to this list. Use it to silence a category, mute a custom term, or allowlist a specific value.
  • Remove entry — Remove this entry from the list, so the category, term or value it covers is detected normally again.
  • Filter by kind — Show only model packs of this kind in the list below. All clears the filter and lists every pack.
  • Open model folder — Open the on-disk folder where packs of this kind are stored, so you can see exactly what is installed and where downloads land.
  • Use this model — Make this installed pack the active model for its kind. The app switches to it on the next run that needs that kind.
  • Remove this model — Delete this pack from disk to free space. You can download it again later from the catalog.
  • Download this model — Fetch this pack from the catalog and install it to disk. Progress shows on the card while it downloads.
  • Refresh detected parsers — Re-check which parsers the current paths would register and update the table below.
  • Enable plugin system — Master switch for plugins. When off, no plugin loads and none of their parsers, actions or hooks are available.
  • Auto-discover on startup — Scan the plugins folder each time the app starts and load whatever is found. When off, you reload plugins manually.
  • Open plugins folder — Open the folder where plugins are installed in your file manager, so you can add or remove them by hand.
  • Reload plugins — Re-scan the plugins folder now and apply any added, removed or changed plugins without restarting the app.
  • Enable Chat AI — Turn on the in-app chat assistant that turns plain-language requests into searches. When off, the AI button is hidden.
  • Browse LLM catalog — Jump to the Local Models tab filtered to LLM packs, where the embedded chat provider picks its model.
  • Anonymize before cloud chat — Mask personal data in your message before it is sent to a cloud model. Strongly recommended when using a cloud provider.
  • AI mode — Choose where the assistant runs: a cloud provider with your own key, a local server you host, or a model embedded in the app fully offline.
  • Manage cloud profiles — Open the Cloud Models tab, where you add and edit the BYOK profiles (provider, key, model) that chat uses.
  • Refresh model list — Re-scan the model folders for gguf files and update the list below with anything added or removed.
  • Select model — Pick this gguf model as the one the embedded provider loads. The selected row is marked with a filled dot.
  • Add model file — Pick a gguf file from disk to add to the model list. Useful for models you downloaded yourself outside the catalog.
  • Open models folder — Open the folder where downloaded gguf models live in your file manager. Drop new files here to have them discovered.
  • Remove custom path — Drop this custom model path from the list. The file on disk is not deleted, only the entry that points at it.
  • Automatic GPU — Let Colpan offload as many model layers to the GPU as it can, then run the rest on CPU. The recommended choice.
  • CPU only — Run the model entirely on the CPU. Useful when the GPU is busy with other apps or you want to save video memory.
  • Manual GPU layers — Set the number of model layers to offload to the GPU by hand. A field appears below for the count. For advanced tuning.
  • Bubble density — Choose how roomy chat bubbles are. Compact tightens them, Cozy adds breathing room, Comfortable is the default.
  • Advanced overrides — Expand or collapse the per-bubble font, line-height and padding overrides that fine-tune the density preset.
  • Reset overrides — Clear the font, line-height and padding overrides so bubbles follow the density preset again. The preset itself is left untouched.
  • Enable Vector DB — Turn on semantic search. This starts a background Qdrant service and downloads an embedding model on first use. Read the warning above first.
  • Auto-start Qdrant — Launch the Qdrant service in the background every time the app starts. When off, you start it by hand from the buttons below.
  • Start service — Start the Qdrant background service now so semantic search can run.
  • Stop service — Stop the Qdrant background service to free its memory. Semantic search is unavailable until you start it again.
  • Refresh status — Re-check whether the Qdrant service is running and update the status shown to its left.
  • Embedding model — Pick the model that turns text into vectors for semantic search. Larger models are more accurate but slower and heavier. Changing it invalidates transferred collections.
  • Manage embedding models — Jump to the Local Models tab filtered to embedding packs, where you download or remove individual models.
  • Hybrid mode — Decide how keyword and vector search combine. Keyword-only uses vectors only on demand; vector-only sends every query to Qdrant; RRF fuses both lanes.
  • Embedder lifecycle — Control when the embedding model is kept in memory. Always-on is fastest but holds RAM; idle-release frees it after a timeout; drop-after-transfer releases it once a transfer finishes.
  • Start on sign-in — Launch Colpan automatically every time you sign in to this computer. When off, you open it yourself.
  • Start hidden in tray — When auto-start is on, launch the app minimised to the system tray instead of opening a window. Enabled only while sign-in start is on.
  • Close to tray — Make the window close button hide Colpan in the tray instead of quitting it. Use the tray icon’s Exit menu to actually quit.
  • Reset to defaults — Restore every field in the section you are viewing to its default value. A confirmation prompt appears first.
  • Cancel reset — Close this prompt without changing anything. Your settings stay as they are.
  • Confirm reset — Replace every field in this section with its default value. This cannot be undone from the interface.

Status Bar

  • Downloads — Model packs currently being fetched, with overall progress. Click to open the downloads panel.
  • Info tips level — Cycle the help-tooltip level: Off, Standard, then Detailed. Detailed adds these explanatory tips on every control.
  • Resource monitor — Live process memory and the number of models and sessions held in memory. Click to open the resource monitor.
  • Indexing progress — Live progress while documents are being indexed. Click to reopen the detailed indexing window.
  • File menu — Open the File menu: import and export indexes, open the config folder, and quit the app.
  • Edit menu — Open the Edit menu: copy, paste and select-all actions for the active input.
  • View menu — Open the View menu: toggle panels, the layout mode and other display options.
  • Indexes menu — Open the Indexes menu: build, rebuild and manage the indexes in the active profile.
  • Tools menu — Open the Tools menu: utilities such as the database viewer and other maintenance actions.
  • PII menu — Open the PII menu: scan for personal data and manage the encrypted PII vault.
  • Settings menu — Open the Settings menu: change app preferences, parsers, theme and language.
  • Help menu — Open the Help menu: documentation, keyboard shortcuts and about information.
  • Minimize — Minimize the window to the taskbar.
  • Maximize or restore — Maximize the window to fill the screen, or restore it to its previous size.
  • Close — Close the window and exit the application.
  • Title bar — Drag this bar to move the window. Double-click to maximize or restore it.
  • New profile — Create a new profile tab. Each profile keeps its own folders, indexes and filters.
  • Profile tab — Switch to this profile. Each profile has its own folders, indexes and search filters.
  • Close profile — Close this profile tab. The profile and its indexes stay on disk; only the tab is removed.
  • Extension filter — Expand or collapse the file-type filter row, where you pick which extensions the results include.
  • Filter preset — Pick a saved file-type preset to apply to the results. Choose No Filter to show every type.
  • Manage filters — Open the filter manager to create, edit and delete file-type presets.