Welcome to MTF Explorer
MTF Explorer is a dual-pane file manager for Windows that lets you work with local drives, network shares, cloud storage, FTP servers, and mobile devices side-by-side in a single window.
Interface Layout
Quick Start
- Use the Drive selector at the top of each pane to pick a drive or connection.
- Click folders to navigate, or type a path directly in the Path Bar. Press Enter to confirm.
- Select files with a click (hold Ctrl or Shift for multiple). Press Ctrl+A to select all.
- Use the toolbar buttons or right-click context menu to Copy, Move, Delete, or Rename.
- Open Settings (top-right) to configure default paths, editor, theme, and language.
File Operations
Copy & Move
Select items in one pane and click Copy (F7) or Move (F8). The destination is always the folder currently open in the opposite pane. A progress overlay appears with per-file and overall progress, speed, and estimated time remaining.
- Verification — if enabled in Settings, SHA-256 checksums are compared after each transfer.
- Conflict resolution — when a file already exists at the destination, a dialog asks whether to skip, overwrite, rename, or apply the same choice to all conflicts.
- Retry on error — transient I/O errors are retried automatically the configured number of times before prompting.
- Cancel — click the Cancel button in the overlay to abort. Already-transferred files are not deleted on cancel.
Drag & Drop
Files and folders can be dragged between the two panes or dropped in from Windows Explorer.
- Drag from one pane to the other to copy the selection to the destination folder.
- Hold Shift while dropping to move instead of copy.
- Drop files or folders dragged from Windows Explorer directly into either pane to copy them into the currently open folder.
- Dropping onto the same pane is ignored — only cross-pane drops are accepted.
Delete
Press Delete or click the toolbar Delete button to send selected items to the Recycle Bin. A confirmation dialog appears before any deletion. To permanently delete without the Recycle Bin, use Shift+Delete.
Secure Delete Premium
Secure Delete performs a 3-pass DoD 5220.22-M overwrite (zeros, ones, random bytes) before removing each file, making recovery significantly harder with conventional tools. Available via the toolbar button or the right-click menu under Delete > Secure Delete.
Rename
Press F2 or click a selected filename after a short pause to start inline renaming. The filename is selected automatically (excluding extension). Press Enter to confirm or Esc to cancel.
Right-click a selection for batch rename modes:
| Mode | Description |
|---|---|
| Simple Rename | Rename a single item inline. |
| Add Prefix | Prepend a string to the names of all selected items. |
| Add Suffix | Append a string before the extension of each selected item. |
| Sequential | Rename items to a base name plus an incrementing counter, e.g. Photo_001.jpg. |
| Regex Replace | Apply a regular expression find-and-replace to all selected names at once. |
| Template | Build new names from a template with placeholders for index, date, original name, and extension. |
New Folder
Click the New Folder toolbar button to create a new directory in the active pane. The folder name is immediately available for inline editing.
Open / View / Edit
- Open (Enter or double-click) — launches the file with the default Windows application.
- View (F5) — same as Open, uses the default associated application.
- Edit (F6) — opens the file in the configured text editor (Notepad, VS Code, Notepad++, or custom). Set your editor in Settings > Editor.
- Print (Ctrl+P) — sends the selected file to the Windows print dialog.
Clipboard Operations
- Copy Name — copies just the filename (without path) to the clipboard.
- Copy Full Path — copies the complete absolute path to the clipboard.
Shell & Terminal
Right-click in any pane to open a terminal or file manager at the current path:
- Open in Explorer
- Open CMD
- Open CMD as Administrator
- Open PowerShell
- Open PowerShell as Administrator
Custom Tools
Custom Tools let you launch any external program or script directly from the MTF Explorer toolbar, with the current pane context automatically injected as arguments. Open the manager via the toolbar wrench icon or through Settings > Custom Tools.
Tool Fields
| Field | Description |
|---|---|
| Name | Label shown on the toolbar button and in the context menu. |
| Executable path | Full path to the program, script, or batch file to run. Use the Browse button to pick a file. |
| Arguments | Command-line arguments passed to the executable. Supports placeholders (see below). |
| Run as Administrator | Launches the tool with elevated privileges via UAC prompt. |
| Show window | When on, the tool opens in a visible window. Turn off for background scripts that produce no UI. |
Argument Placeholders
Use these placeholders in the Arguments field — they are replaced at launch time with live pane values:
| Placeholder | Replaced with |
|---|---|
{path} | Full path of the currently active pane folder. |
{left} | Full path of the left pane folder. |
{right} | Full path of the right pane folder. |
{selected} | Space-separated quoted full paths of all selected items in the active pane. Empty string if nothing is selected. |
{name} | File or folder name (without path) of the first selected item. |
code and Arguments to {path}. To diff two panes with WinMerge: WinMergeU.exe with Arguments "{left}" "{right}".Accessing Custom Tools
- Each tool appears as a button in the top toolbar immediately to the left of the manage-tools button.
- Tools also appear in the right-click context menu under the Custom Tools submenu (if that item is enabled in the context menu editor).
- Click the manage-tools button (wrench icon) to add, edit, reorder, or remove tools at any time.
Context Menu Customization
The right-click context menu is fully customizable. Open Settings > Context Menu to reorder, hide, or group menu items to match your workflow.
Available Actions
Every built-in action can be individually shown or hidden and placed anywhere in the menu. Available actions include:
- Open, Open With, Open in Explorer
- Open Terminal (CMD, CMD as Admin, PowerShell, PowerShell as Admin)
- Copy, Move, Send To
- Open Archive, Zip, Unzip
- New Folder, New Text File, Convert Image, Rename
- Copy Name, Copy Full Path
- Delete (including Secure Delete)
- Tags
- Properties, System Properties
Using the Editor
- Drag items up or down to reorder them in the menu.
- Add separators between items to create visual groups.
- Add a custom group (submenu) with a name and icon, then drag actions into it to nest them one level deep.
- Toggle the visibility checkbox next to any item to hide it without removing it from the layout.
- Click Reset to Default to restore the original factory layout at any time.
View & Display
Thumbnails
Toggle the Thumbnails button in the toolbar to show image previews alongside filenames. Works for common image formats (JPEG, PNG, BMP, GIF, WebP, HEIC) as well as video thumbnail frames. Thumbnail loading is asynchronous and does not block scrolling.
Preview Panel
Click the Preview toolbar toggle to open a side panel showing a larger thumbnail, the file name, type, size, dimensions (for images/video), and modification date. Drag the splitter between the file list and preview panel to resize.
Column View (Miller Columns)
Toggle the Columns view button (next to the Thumbnails button) to switch to a macOS Finder-style column layout. Each column shows the contents of a folder. Clicking a subfolder immediately opens it as an adjacent column to the right, letting you navigate a directory tree without leaving the current view.
- Single-click a folder to expand it into the next column. Columns to the right are automatically removed.
- Single-click a file to select it. Double-click to open it with the default application.
- The path bar updates as you navigate deeper into the tree.
- The column view mode is saved per pane and restored on next launch.
Custom Columns
Click the Columns button to toggle optional metadata columns:
| Column | Source | Applicable Formats |
|---|---|---|
| Dimensions | Width × Height in pixels | Images, Videos |
| Date Taken | EXIF DateTimeOriginal | JPEG, HEIC, RAW |
| Duration | Media stream length | Video, Audio |
| Media Title | ID3 / WMA tag | MP3, FLAC, WMA |
| Artist | ID3 / WMA tag | MP3, FLAC, WMA |
Sorting
Click any column header to sort by that field. Click again to toggle ascending/descending order. The current sort direction is shown with an arrow indicator. Sort state is saved per workspace tab.
Available sort fields: Name, Size, Date Modified, Attributes, and any visible custom columns.
UI Density
In Settings > UI Density, switch between Normal and Compact modes. Compact mode reduces row height and font sizes for more items visible at once.
Inline Tag Display
Click the Tags toolbar toggle to show tag chips beneath each filename. Click a chip to open the tag picker for that file, or press Ctrl+T on a selected item.
Search
MTF Explorer provides three complementary search mechanisms: filename pattern search (instant, no index required), a local filename index (fast, SQLite-backed), and AI semantic search (natural language, embedding-based). All three are accessed from the Search button in each pane's toolbar.
Filename Pattern Search
Type a pattern in the Pattern box. The search starts from the folder currently open in that pane.
- * matches any sequence of characters — e.g.
report*.pdf - ? matches exactly one character — e.g.
file?.txt - Enable Regex to use full .NET regular expressions — e.g.
^invoice_\d{4}\.xlsx$
The Extensions box narrows results to specific file types: enter a semicolon-separated list such as .jpg;.png;.webp.
Advanced Filters
Expand the Advanced panel for additional filter controls:
| Filter | Description |
|---|---|
| Include Subfolders | Search recursively inside all child directories (on by default). |
| Hidden Files | Include files and folders with the Windows Hidden attribute. |
| System Files | Include files with the Windows System attribute. |
| Date Range | Filter by last modification date using From / To calendar pickers. |
| Size Range | Minimum and maximum file size with unit suffixes: 100KB, 10MB, 1GB. |
| Required Tags | File must have all listed tags (comma-separated). |
| Excluded Tags | File must have none of the listed tags. |
Duplicate Detection
Enable Find Duplicates to locate files with identical content. Enable Verify Hashes to use SHA-256 comparison for byte-perfect accuracy (slower but authoritative). Results are grouped so each duplicate cluster appears together in the list.
Local Search Index
The local search index is a SQLite database of filenames built ahead of time, making pattern searches across large directory trees nearly instant. It is separate from AI semantic search and operates entirely on filenames — not file contents.
Configuring the Index (Settings > Search Index)
| Setting | Description |
|---|---|
| Enable | Turn the index on or off. When off, pattern searches fall back to live directory enumeration. |
| Auto-refresh | A file system watcher keeps the index current as files are added, renamed, or deleted. Recommended for drives that change frequently. |
| Index Roots | The top-level folders to include. Click Use Current to add the active pane path. You can add multiple roots across different drives. |
| Rebuild | Triggers a full re-scan of all roots. Use when the stale warning appears or after adding a new root. |
| Clear | Deletes the index database entirely. A rebuild is required before fast search is available again. |
AI Semantic Search Premium
AI semantic search embeds both your files and your query as high-dimensional vectors and returns results by similarity — not keyword matching. You can describe what you are looking for in plain language and find files whose names, folder paths, or contents capture that meaning, even if none of the exact words appear.
Each search combines two complementary techniques via Reciprocal Rank Fusion: a dense semantic pass (embedding similarity) that understands meaning, and a BM25 keyword pass (full-text search) that excels at exact terms, version numbers, and identifiers. The two ranked lists are merged so results strong in either dimension float to the top.
All processing runs entirely on your device. Models are downloaded once from Hugging Face and stored locally. No filenames, paths, or content ever leave your machine.
Quick Start
- Open Settings > AI Search and enable AI Search.
- Pick a model (see the Model Reference table below). Click Download and wait.
- Add one or more Index Roots (the folders you want searchable).
- Click Build Index. A progress bar shows files processed. After the build, a file system watcher keeps the index up to date automatically as files change.
- In the search panel, switch on AI Search, type a natural-language query, and click Search.
Model Reference
Choose the model that best fits your balance of speed, disk space, and query quality. All models listed below are open-source (Apache 2.0 or MIT) and run on CPU via ONNX Runtime — no GPU required.
| Model | Size | Dims | Best For |
|---|---|---|---|
| MiniLM-L6 | ~22 MB | 384 | Fastest indexing and query time. Great starting point for filename-only search on any hardware. Apache 2.0. |
| MiniLM-L12 | ~33 MB | 384 | Slightly better recall than L6 with minimal extra overhead. Still very fast. Apache 2.0. |
| BGE-small-en | ~130 MB | 384 | Higher-quality English-language embeddings. Good recall on complex or abstract queries. MIT. |
| BGE-base-en | ~440 MB | 768 | Strong English quality with 768-dimensional vectors. Recommended for content-heavy indexes. MIT. |
| MPNet-base | ~438 MB | 768 | General-purpose sentence embeddings. Handles varied query styles (questions, descriptions, tags) well. Apache 2.0. |
| LaBSE Multilingual | ~471 MB | 768 | Supports 109 languages. Use when your files or queries mix languages. Apache 2.0. |
| BGE-large-en | ~1.3 GB | 1024 | Best English embedding quality available. Highest recall on nuanced queries. Requires substantial disk and RAM. MIT. |
| CLIP ViT-B/32 | ~350 MB | 512 | Image search. Indexes images by visual content — type what you see to find it. Text-to-image only; does not read filenames or OCR. Requires the Images file type group. Apache 2.0. |
Indexed File Types & Content Extraction
When Index file contents is enabled, the indexer streams the full text of each file and splits it into overlapping chunks — no file size limit applies to text and code files. Large source files, logs, and documents are fully covered regardless of size. Only the size of the embedding model's context window limits how much fits in each chunk.
| Category | Extensions | How content is extracted | Limits |
|---|---|---|---|
.pdf |
Text is extracted page-by-page using PdfPig with spatial word reordering (top-to-bottom, left-to-right), correcting the arbitrary stream order produced by InDesign, LaTeX, and similar tools. Document metadata (title, subject, author) is prepended to the first chunk. Each chunk is split at sentence boundaries to preserve sentence context. | Up to 200 pages per file. Encrypted or image-only PDFs fall back to filename-only indexing. | |
| Markdown | .md |
H1, H2, and H3 headings are collected and placed at the start of the first chunk as a structural summary. The remaining body prose is then streamed and chunked in order. | No size limit. Full file indexed. |
| JSON | .json |
Top-level key names (the schema surface) are extracted and placed at the start of the first chunk — key names like invoiceNumber, customerId, lineItems are far more semantically dense than raw values. The full file content follows as streamed chunks. | No size limit. Full file indexed. |
| YAML / TOML | .yaml .yml .toml |
Section headers and top-level key names are extracted as a summary prefix. The full file content is then streamed and chunked. | No size limit. Full file indexed. |
| Code | .cs .py .js .ts .go .java .cpp .c .h .hpp .rs .rb .php .swift .kt |
Class, function, interface, struct, and method names are extracted via language-specific patterns and placed at the start of the first chunk, giving the embedding a strong semantic fingerprint even for large files. The full source code is then streamed in overlapping chunks covering every line. camelCase and PascalCase identifiers are automatically expanded in the keyword index — searching for "connection pool timeout" will match connectionPoolTimeout, ConnectionPoolTimeout, and similar identifiers. | No size limit. Full file indexed. |
| Documents / Data / Web / Scripts | .txt .csv .log .xml .ini .cfg .conf .sql .html .htm .css .sh .bat .ps1 |
Read as plain text; whitespace is collapsed. The full file is streamed and split into overlapping chunks at sentence boundaries where possible. | No size limit. Full file indexed. |
| Images (OCR) | .jpg .jpeg .png .bmp .tiff .tif .gif .webp |
Windows built-in OCR extracts visible text from the image. Very large images are scaled down to 2 000 px on the long side before recognition. The recognized text is embedded as a single chunk alongside the filename. | Max 20 MB per image. Files exceeding this size are indexed by filename only. OCR requires a Windows language pack to be installed. |
| Images (CLIP) | .jpg .jpeg .png .bmp .tiff .tif .gif .webp |
When using the CLIP ViT-B/32 model, images are encoded by their visual appearance using OpenAI's vision encoder at 224x224 pixels — not OCR text. Type what you see: "sunset over mountains", "spreadsheet with blue header", "cat sleeping on keyboard". | No explicit size limit. Images are resized to 224x224 before encoding. |
Content Chunking Settings
Text and code files are streamed and split into overlapping windows. A sliding buffer of at most two chunk-widths is held in memory at once, so even very large files are processed without loading the entire content into RAM. PDFs are chunked per page using the same sentence-boundary splitter.
| Setting | Default | Description |
|---|---|---|
| Chunk size | 1 000 characters | Maximum number of characters in each chunk. Range: 200-3 000. Roughly 200-750 tokens depending on content language. Larger chunks capture more context per embedding but may dilute specificity for precise queries. The embedding model's hard limit is 512 tokens; content beyond that is silently truncated by the model, so keeping chunks at or below 2 000 characters ensures full coverage. |
| Chunk overlap | 150 characters | Number of characters repeated at the start of each successive chunk. Range: 0-600. Overlap prevents a concept that straddles a chunk boundary from being missed. Higher overlap produces more chunks (slower build, larger index) but improves recall on long documents. |
AI Search Settings Reference (Settings > AI Search)
| Setting | Description |
|---|---|
| Enable AI Search | Master toggle. When off, the AI search toggle in the search panel is hidden. |
| Model | Select an embedding model from the dropdown. A short description and download size are shown below the picker. |
| Download | Downloads the selected model's ONNX weights and vocabulary from Hugging Face. A progress bar shows download progress. The files are stored in the application data folder. |
| Index Roots | Folders to scan and index. Click Use Current to add the active pane path. Multiple roots are supported. |
| Index file contents | When enabled, supported file types are opened and their full text content is embedded alongside filename context. Significantly increases index size and build time but enables document-level and code-level semantic search. |
| Chunk size | Characters per content chunk (200-3 000, default 1 000). See Chunking Settings above. |
| Chunk overlap | Overlap characters between chunks (0-600, default 150). See Chunking Settings above. |
| File types | Checklist of extension groups. Deselect categories you do not need to keep the index smaller and build faster. For CLIP, only the Images group is relevant. |
| Build Index / Rebuild | Starts (or restarts) a full index build over all configured roots. Progress is shown as a file count. After completion, a file system watcher keeps the index up to date automatically as files are added, changed, renamed, or deleted. |
| Cancel | Stops an in-progress index build. The partially built index is still usable for already-processed files. |
| Clear | Deletes the entire AI index for the selected model. The model download is not affected. |
| Status / Diagnostics | Shows the number of indexed vectors, the last build time, and any errors encountered during the last build. |
Query-Time Controls (Search panel)
| Control | Description |
|---|---|
| Max Results | Cap on the number of results returned (1-500, default 50). Higher values surface more candidates but include lower-confidence matches. |
| Min Score | Similarity threshold (0-100%). Results below this score are filtered out. The score reflects cosine similarity for semantic matches; keyword-only matches (found by BM25 but not semantic search) receive a smaller score decaying by rank. Raise the threshold to see only high-confidence semantic matches; lower it if you are getting too few results. |
CLIP Image Search in Detail
CLIP (Contrastive Language-Image Pretraining) is fundamentally different from the text embedding models. Where text models find files whose names or content match your query semantically, CLIP finds images whose visual appearance matches your text description.
- Encoding — during indexing each image is run through the CLIP vision encoder and stored as a 512-dimensional vector.
- Querying — your text query is encoded by the CLIP text encoder into the same 512-dimensional space. Results are ranked by cosine similarity between the text vector and each image vector.
- What works well — visual descriptions: "screenshot with dark background", "handwritten notes", "bar chart", "dog playing in snow".
- What does not work — CLIP does not read text inside images (no OCR). Do not use it to find images containing a specific word; use the OCR-capable text models for that.
- File types — only the Images file type group is indexed when CLIP is selected. Enabling other groups has no effect.
- Setup — CLIP downloads two ONNX files (text encoder + vision encoder) totalling ~350 MB.
Getting the Most Out of AI Search
- Start with MiniLM-L6 for a low-friction first experience. Upgrade to BGE-base or MPNet if recall is poor on specific query types.
- Use descriptive phrases, not single keywords. "quarterly sales report Europe 2023" outperforms "report".
- Enable file content indexing for document-heavy or code-heavy folders. Without it, only filenames and paths are matched — the full text of PDFs, source files, and documents is invisible to the search engine.
- Search code by concept, not exact identifier. Because camelCase identifiers are automatically broken into words in the keyword index, searching "connection pool timeout" will match files containing connectionPoolTimeout, ConnectionPoolTimeoutException, and similar names.
- Keep chunk overlap at 150+ for PDFs and long text files to avoid losing information at chunk boundaries.
- Lower Min Score (to 20-30%) if you get zero or very few results. The right file may be present but scored below the default threshold. Keyword-only matches (found by BM25) appear with a small non-zero score even if semantic similarity is low.
- Raise Min Score (to 70-80%) when results are noisy and you know exactly what you want.
- Limit file type groups to what you actually search. Disabling large code or log groups cuts build time and index size significantly.
- Use LaBSE Multilingual if your files or queries span multiple languages — all other models are English-optimised.
- For image libraries, use CLIP with concise visual descriptions. For images that contain text you want to search (screenshots, scanned docs), use any text model with the OCR-capable Images group.
- Rebuild after major file reorganisations. The auto-watcher handles incremental changes (creates, edits, deletes, renames) but a full rebuild ensures nothing is missed after large batch moves or copies.
Find Similar Files
Right-click any file and choose Find Similar Files (AI) to find other files in the same pane whose content or name is semantically close to the selected file. Results replace the pane's file list and are scored by similarity, just like a regular AI search.
- Requires AI Search to be enabled and an index to be built for the current pane's folder.
- The search is scoped to the folder currently open in the pane — files outside that folder are not returned even if they are indexed.
- The embedding used is the centroid (average) of all content chunks stored for the selected file, giving a whole-file representation rather than matching on a single excerpt.
- If the selected file has not been indexed yet, it is embedded on the fly using the active model before searching.
- Only available for files, not folders.
Disk Usage Analysis
The Disk Usage panel scans the current folder and displays a Squarified Treemap — a rectangle-based visualization where each block's size corresponds to the disk space consumed by that item. Larger blocks represent larger folders or files.
Opening the Panel
Click the Disk Usage button (treemap icon) in the left or right pane toolbar. The panel opens inline above the file list for that pane and begins scanning immediately. A refresh indicator appears during the scan.
Reading the Treemap
- Rectangle size is proportional to total folder size (all contents, recursive).
- Color depth indicates folder nesting level; darker shades are deeper.
- Each rectangle shows the folder name and size when large enough to fit.
- Hover a block to see the full path and exact size in the footer.
Navigating the Treemap
- Double-click a block to drill down into that folder (treemap re-draws for that subtree).
- Use the breadcrumb in the footer to navigate back up the hierarchy.
- Right-click a block and choose Explore content to navigate the parent pane to that folder.
Full-Screen Mode
Click the Expand button (⤢) in the disk usage panel header to switch to a full-screen overlay that spans the entire application window. This gives much more space for large folder trees. Press the Close button or Esc to return to normal view.
Refreshing
Click the Refresh button in the panel header to re-scan the folder from scratch. This is useful after copying, deleting, or moving files.
Archive Operations
Create ZIP
Select one or more files or folders, then click the Zip (compress) toolbar button or choose Create ZIP from the context menu. A dialog lets you name the archive; it is created in the same folder as the selected items. If an archive with that name already exists, you are asked to overwrite or rename.
Extract ZIP
Select a .zip file and click the Unzip (extract) toolbar button or choose Extract ZIP from the context menu. Contents are extracted into a subfolder named after the archive. If files already exist at the destination, the conflict resolution dialog appears.
Browse Archive as Folder
Select a .zip and choose Open Archive as Folder from the context menu. The archive is mounted as a virtual folder; you can navigate inside, copy files out, and view contents without extracting. To leave the archive, click the Up button or navigate to a different path.
Cloud & Remote Connections Premium
Click the Connections button in the top bar to manage all remote connections. Once configured, each connection appears in the Drive selector of both panes and can be browsed, copied to/from, and used in search.
Google Drive
Provide a label, a Client ID, a Client Secret (from the Google Cloud Console), and optionally a Shared Drive ID for Google Workspace shared drives. After saving, an OAuth browser window appears to authorise access. Credentials are stored in the Windows Credential Manager and not included in settings exports.
Azure Blob Storage
Provide a label, the connection string from your Azure Storage account (found in Settings > Access Keys in the Azure Portal), and the target container name. Files appear as a flat list (Azure Blob does not have real folders; the path separator is emulated).
AWS S3
Provide a label, AWS profile or access key/secret, region, and bucket name. You can optionally specify a prefix to browse only a subdirectory of the bucket. Works with any S3-compatible storage service.
FTP / SFTP / FTPS
Provide a label, host, port (default 21 for FTP, 22 for SFTP), username and password, and select the protocol:
- FTP — plain (unencrypted)
- FTPS — FTP with TLS (implicit or explicit)
- SFTP — SSH File Transfer Protocol (can use a private key file instead of password)
MTP Devices (Phones, Cameras)
Connect a phone, camera, or other MTP/PTP device via USB. Open Connections > Manage MTP to see detected devices. Select a device and click Browse to load it in a pane — the device appears in the drive selector. You can copy files from the device to a local folder. Uploading to MTP devices is not supported due to Windows API limitations.
Compare & Sync
Click the Compare button in the top bar to open the comparison overlay. It analyses the contents of the left and right panes side-by-side and shows synchronisation status for each item.
Filter Views
Use the Filter dropdown to narrow the displayed items:
Synchronisation Operations
| Operation | Description |
|---|---|
| Sync Selected | Copy checked items to whichever side is missing or older. |
| Sync Left → Right | Copy all items from the left pane to the right, overwriting older files. |
| Sync Right → Left | Copy all items from the right pane to the left, overwriting older files. |
| Resolve Conflicts | Open each conflicting file pair and choose the resolution manually. |
Refreshing
Click Refresh in the compare overlay to re-scan both panes and update all status indicators. Do this after a sync operation or after modifying files externally.
Workspaces
Workspaces are named tabs that each remember an independent pair of left and right paths, plus sort state (field and direction) for each pane. Switch between them instantly from the dropdown in the top bar.
Managing Workspaces
- Add — click the + button next to the workspace dropdown to create a new workspace starting from the current paths.
- Rename — click the workspace name in the dropdown list to edit it inline.
- Remove — hover a workspace entry and click the remove icon. At least one workspace is always kept.
- Switch — click any workspace entry in the dropdown. Both panes and sort states update instantly.
Persisted State
Each workspace remembers:
- Left and right folder paths
- Sort field and direction for each pane
- Active custom column visibility
State is saved automatically when you navigate away or close the application.
Settings
Open Settings from the gear icon in the top-right toolbar. Settings are organised into the following sections:
Default Paths
Set the default left and right pane paths that are loaded when the application starts (or when a new workspace is created). Use the browse button to pick a folder. Leave blank to use the last-used path.
Editor
Choose which application is launched when you press F6 (Edit):
- Notepad (Windows built-in)
- Visual Studio Code
- Notepad++
- Custom — enter the full path to any executable
Transfer Settings
| Setting | Description |
|---|---|
| Progress Update Interval | How often (in ms) the progress overlay refreshes during a transfer. Lower values are more responsive but use more CPU. |
| Verify Transfers | Compare SHA-256 checksums after each copy operation. Slower but detects silent corruption. |
| Show Progress List | Display individual file entries in the copy overlay alongside the overall bar. |
| Retry Attempts | Number of times to retry a failed I/O operation before prompting the user (1–10). |
| Buffer Size | Read/write buffer size in MB for file copy operations (64–512 MB). Larger buffers improve throughput on fast drives. |
| Calculate Folder Sizes | Automatically compute the total size of each visible folder. May slow navigation in very large directories. |
UI & Appearance
| Setting | Description |
|---|---|
| Theme | System (follows Windows setting), Light, or Dark. |
| UI Density | Normal or Compact. Compact reduces row heights and font sizes. |
| Language | Select the display language. Supported: English, German, French, Spanish, Romanian, Portuguese (BR). |
| Minimize to Tray | When enabled, closing the window sends the app to the system tray instead of exiting. |
Data Management
- Export Settings — saves all settings to a
.jsonfile. You can choose to include tags, auto-tag rules, and context menu layout. Cloud credentials are never exported. - Import Settings — load settings from a previously exported file. A confirmation dialog shows what will be overwritten.
- View Logs — opens the application log folder in Windows Explorer for troubleshooting.
Keyboard Shortcuts
All shortcuts below are the defaults. To customise them, open Help > Keyboard Shortcuts. Conflicts are detected automatically and highlighted in the shortcut editor.
Navigation
File Operations
Shell & Terminal
Tags
Subscription & Licensing
Trial Mode
MTF Explorer starts in Trial mode, which provides full access to core features for a limited period. The subscription chip in the top bar shows the remaining trial days. Core features available in trial:
- Dual-pane navigation, copy, move, delete, rename
- Thumbnails, column view, preview panel, custom columns, sorting
- Advanced search with filters and duplicate detection
- ZIP create and extract
- Tags and auto-tag rules
- Multiple workspaces
- Compare and sync
- Disk usage analysis
Premium Features
- Cloud connectors (Google Drive, Azure Blob, AWS S3)
- FTP / SFTP / FTPS connections
- MTP device browsing (phones, cameras)
- AI semantic search
- Secure Delete (3-pass DoD overwrite)
Activating a License
Click the subscription chip in the top bar (labelled Trial or License required) and follow the prompts to subscribe or enter a license key. An active internet connection is required for activation. Once activated, the chip changes to Licensed.
Managing Your Subscription
Click the Licensed chip and choose Manage to view your subscription details, renewal date, and billing options through the subscription portal.