Complete Tutorial and Reference
Learn how to use FilePulse effectively with our comprehensive guide covering basic operations to advanced configurations.
How to run FilePulse for the first time
FilePulse can run on Windows, macOS, and Linux. The default local web console uses ports 80 and 443; if these ports are occupied or your environment needs a different network setup, adjust config.json before startup.
To change ports, modify the config.json file
On Windows, run FilePulse.exe with administrator privileges when disk indexing, file operations, or low-port binding requires it. On macOS and Linux, follow the package-specific startup instructions from the download page.
FilePulse.exeOpen your browser and navigate to http://127.0.0.1, then enter the default username and password (both are "root") to access the system
http://127.0.0.1Username: root, Password: rootIf FilePulse.exe displays a shield icon after download, please move it to another folder until the shield disappears, otherwise it will affect opening a target with normal permissions. You can ignore this if you don't need this feature.
If running normally, the program will automatically switch to administrator privileges, but please do not set "Run this program as administrator" in the compatibility tab of FilePulse.exe properties. If compatibility is set, it will have the same issue as mentioned above and cannot open a target with normal permissions. You can ignore this if you don't need this feature.
If "Volume change log is inactive" appears, it means there is a problem with a drive letter. You need to modify config.json to remove the problematic drive letter from drive_list.
After version upgrade, it is recommended to place the newly downloaded FilePulse in a separate new folder to avoid conflicts between old and new configurations.
The software currently uses packing protection, which may cause false positives from antivirus software. Please add FilePulse.exe to the whitelist to resolve this issue.
Explore the main capabilities of FilePulse
Similar to Everything, with various complex filtering conditions, mainly used for users' complex searches, can respond in milliseconds
Search inside text, Office documents, PDFs, books, web pages, and image-based content with OCR. Text inside screenshots, scanned PDFs, and embedded document images can become searchable after indexing.
Use an image, screenshot, or video as the query to find visually similar images and matching video segments, then review thumbnails, paths, servers, and target time ranges.
Similar to Listary, input keywords and results are pre-sorted according to most habits, quickly locate target location and launch
Compare files by size and signatures to locate duplicates quickly, review results safely, and clean up storage efficiently.
Remote quick preview for documents, books, archives, images, audio, and video, plus download, upload, and remote file location.
Continuously sync mapped local and remote folders with periodic updates, reconnect resilience, and clear deletion tracking.
Manage LLM options, keep persistent multi-session history, support image upload, and clear all sessions in one click.
Similar to Sonic Launcher, pre-add commonly used software (due to browser limitations, drag and drop cannot get complete paths, currently requires manual input, future native app will support drag and drop)
Uses the core search and remote capabilities; the launcher panel is optional in cluster mode.
Auto-fill the current file manager path with a shortcut across Windows, macOS, and Linux, continue from supported open dialogs on Windows, and refine or step up the active search scope when needed.
Manage ongoing search tasks with persistent annotations, grouping, pinned items, custom sorting, startup restoration, and smoother keyboard navigation.
A clear path from discovery to remote auto-sync, quick preview, AI conversations, cleanup, and delivery.
Use filters, regex, and pinyin matching to locate targets instantly.
Search across document text and OCR-recognized image text with indexing, scoring, highlights, and quick preview.
Upload an image or video, or paste a screenshot, to find similar images and jump to matching video segments.
Open apps and folders fast with habit-based sorting.
Find redundant files and keep only the versions you need.
Upload, download, and locate files across nodes with smooth remote access.
Keep mapped folders between local and remote nodes continuously synchronized with automatic updates.
Instantly preview documents, images, and media without leaving the workflow.
Continue work through contextual AI chats with image upload and persistent session history.
Share selected content with controlled access when collaboration is needed.
Start from the current folder faster, then manually refine the scope or jump to the parent scope as the search expands.
Use persistent annotations, grouping, pinned items, custom sorting, and smoother result navigation to keep complex searches under control.
FilePulse supports content searching across multiple file formats
Different file types are processed by specialized handlers, including OCR handlers for images, scanned PDFs, and pictures embedded in Office documents. Choose handlers by file type in Settings -> Content Rules / Handler Profiles.
Use these handlers when the document already contains selectable text, structured document text, book content, or web page text.
Reads all text-type files as is
All text content from Microsoft Word documents
All text content from Microsoft Excel spreadsheets
All text content from Microsoft PowerPoint presentations
All text content from PDF documents
All text content from Rich Text Format documents
Searchable content from EPUB book files
Searchable content from FB2 book files
HTML web page content with tags removed
Use these handlers when important information appears inside screenshots, photos, scanned pages, image-heavy PDFs, or images embedded in Office documents.
Word document text plus OCR-recognized text from embedded images
Excel spreadsheet text plus OCR-recognized text from embedded images
PowerPoint slide text plus OCR-recognized text from images in slides
PDF text plus OCR-recognized text from scanned pages and embedded images
OCR-recognized text from image files so screenshots and photos can be found by content
Use these file types when images or videos should be available for visual search, screenshot matching, and video segment review. Video visual search requires FFmpeg to be available.
Images and common video files can be used as visual search targets or query files. Image search works directly; video indexing and querying require FFmpeg.
Tip: use OCR handlers when searchable information is stored as an image rather than selectable text.
Master these shortcuts to boost your productivity
Unlock the full potential of FilePulse
The built-in certificate is a development certificate. To replace with a real certificate, replace the files in the certs directory.
HTTP2/HTTP3 are enabled by default, but may automatically downgrade to HTTP2 due to development certificate issues. After replacing with a real certificate, HTTP3 can be used normally.
Search Content and Path edit boxes support NOT (!), AND (space), OR (
Search Content and Path edit boxes support NOT <not>, AND <and>, OR <or>, left parenthesis <left>, right parenthesis <right>
Four search modes: normal mode, regular expression, prefix matching, fuzzy matching. Fuzzy matching has the worst performance, try to avoid using it. Prefix matching requires at least two word segments to start, and in simple expressions it is separated by spaces, so prefix matching needs advanced expressions.
Two search modes: wildcard search, regular expression search. Both perform individual matching on all files to be queried, performance will be worse.
In the login interface, directly enter the address and port number of the remote server in API Base URL, such as https://abc.com:443, then enter the username and password to connect to the remote server.
Add the target server address, port number, username and password in Remote Servers, the system will automatically connect to the target server. Use https:// for http2 and http3 when writing addresses, use http:// for http1.1.
After connection, you need to mount the target server drive to local drive, click Mount Servers, select the server name just connected, and write the address of the server being accessed in the callback address.
Mount takes about a few seconds depending on how many files the target server has. After completion, you can search remote server file names and file contents just like locally. Searching remote file content will transfer large amounts of data, it is recommended to use score mode to only take top10.
When mounting a WebDAV service in Windows, note the following settings for HTTP access and file size limits.
Press Win + R, enter regedit, and open Registry Editor.
Go to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\WebClient\Parameters.
Find BasicAuthLevel and set its value to 2.
Press Win + R, enter services.msc, and open Services.
Find WebClient and restart the service.
The default file size limit for Windows WebDAV transfers is 50 MB.
In the same registry path, find FileSizeLimitInBytes.
Change the default value 2faf080 to ffffffff to raise the maximum transferable size from 50 MB to 4 GB.
Restart the WebClient service.
Before using sharing on Windows, enable WebClient basic authentication in the registry.
Right-click Start and choose Run.
Type regedit and click OK.
In Registry Editor, navigate to HKEY_LOCAL_MACHINE > SYSTEM > CurrentControlSet > Services > WebClient > Parameters.
Open BasicAuthLevel and set Value data to 2.
Restart the computer and use sharing again.
config.json controls the local service identity, login, ports, indexed locations, shortcuts, startup behavior, and a few runtime preferences. Most users only need to adjust ports, drive_list, hot_keys, instance policy, or startup options when the default environment is not suitable.
Ports, drive_list, hot_keys, multi-instance policy, and startup options are reasonable fields to adjust manually. Account credentials should be changed from the web interface whenever possible. Restart FilePulse after changing ports, monitored locations, hotkeys, or max_batch so the new behavior can be applied consistently.
server_nameDisplay name of the local FilePulse node. Keep it unique when you use remote nodes or cluster workflows.
username / password / last_password_changeAdministrator login information and the timestamp used to invalidate older sessions after a password change. Prefer changing username or password in the web interface.
port_http / port_httpsHTTP and HTTPS ports used by the local web console and service APIs. If ports 80 or 443 are occupied, change these values before startup and visit the matching address.
drive_listLocations FilePulse monitors and indexes. On Windows this is usually drive letters, while macOS and Linux use paths such as /Users or /home.
hot_keysGlobal shortcut definitions for current folder search, content search, quick launcher, launcher panel, and context search.
open_link_after_bootControls whether FilePulse opens the browser automatically after startup. Disable it for background startup or server deployments.
max_batchBatch size for content indexing. Larger values may build faster but require more memory and CPU. Lower it if content indexing cannot allocate memory.
hide_consoleControls whether the console window is hidden when FilePulse starts in environments that support it.
native_logControls native-side logging behavior for troubleshooting startup, shortcut, or desktop integration issues.
allow_multi_filepulseControls whether multiple FilePulse service instances can run at the same time.
allow_multi_nativeControls whether multiple native desktop helper instances can run at the same time.
boot_autostartControls startup registration. Use it when you want FilePulse to start automatically after login or boot.
Settings import/export can migrate config.json together with sharing, sync, LLM, content rules, remote server, and search workspace data. This guide only explains config.json because the other files are better managed from the web interface.
If FilePulse closes itself after the machine resumes from sleep or a similar state, this is usually not a program defect. It happens because FilePulse's online verification is interrupted when the terminal enters sleep or a similar state, so the program exits automatically. Simply restart FilePulse.
If memory cannot be allocated when building search content, max_batch is set too large. Just modify the config.json file and set max_batch to a smaller value, then run again.
If search content build gets stuck, you can delete all entries in rules in rules.json, which is equivalent to clearing all rules, then run again.
Build results are in the index directory, which can be deleted at any time to rebuild. I personally recommend not building overly large plain text, such as a 500MB Word document with mostly images will be fine, but if it is a 500MB plain text file with pure text content, the search system may take a very long time to build. And such large plain text files will consume huge amounts of memory during building.