Everything Disk Usage is a Windows desktop disk-usage viewer powered by the voidtools Everything index. It asks Everything for indexed file paths and file-size metadata, then builds an interactive folder-size tree, pie chart, file list, and duplicate-file view without walking the disk from scratch.
The latest rebuilt installer checked into this repository is:
Download EverythingDiskUsage-Setup-1.0.0.exe
The installer is self-contained, so it includes the .NET runtime files needed by the app. You still need Everything Search installed, running, and fully indexed.
- Windows 10/11 x64
- Everything Search installed from voidtools
- Everything Search running in the background with its database loaded
- .NET 11 SDK Preview 6 or newer only if you want to build from source
- Inno Setup 6 only if you want to rebuild the installer locally
- A supported local AI accelerator and internet access for the initial Foundry runtime/model download only if you enable AI advice
Install Inno Setup machine-wide with:
winget install JRSoftware.InnoSetup --scope machine- Download EverythingDiskUsage-Setup-1.0.0.exe.
- Run the installer.
- If the installer warns that Everything Search is missing, install Everything Search before using the app.
- Launch Everything Disk Usage from the Start menu or the optional desktop shortcut.
The installer uses a per-user install location under %LOCALAPPDATA%\Programs\EverythingDiskUsage, so it does not require administrator elevation.
- Make sure Everything Search is open and finished indexing.
- Choose a drive from the drive selector, or type/paste a folder path into the path box.
- Click Scan.
- Watch the progress bar and status text while the Everything SDK query runs.
- Click Cancel if you need to stop an active scan.
The app relies on Everything's existing index, so scans are usually much faster than tools that enumerate the filesystem directly. If Everything is still indexing, missing, or not running, the scan can fail or return incomplete data.
The Chart tab is the main exploration view.
- The left pane shows a folder tree ordered by size.
- The right pane shows the selected folder as a pie chart.
- Click a folder in the tree to update the chart.
- Click a pie slice for a subfolder to drill into that folder.
- Use the back button above the chart to return to the parent folder.
- The legend shows each visible slice with size and percent values.
- Very small or lower-ranked items are grouped into Other so the chart stays readable.
The selected folder summary shows total size, file count, and folder count. Direct files in the selected folder appear as Files in this folder in the chart.
The Details tab gives a table-oriented view of the same scan.
- The folder grid lists every scanned folder with path, size, file count, folder count, percent of parent, last modified, and last accessed values.
- Selecting a folder updates the file grid to show files inside that folder scope.
- File rows are grouped by file name so repeated names are easy to compare.
- Groups with multiple files expand into individual file rows with their actual directories.
- The file grid caps visible rows to keep the UI responsive on very large scans.
Use this tab when you need exact paths, file counts, timestamps, or a sortable table instead of a chart.
The Duplicates tab groups files that have the same file name and the same size.
- Each row is a real file occurrence and shows its location, copy count, per-file size, and estimated wasted space.
- Use Group by to expand duplicate sets by their parent folder, higher ancestors, or drive/root.
- Folder and duplicate-set groups are expandable, so you can compare where copies cluster at the depth that matters to you.
- Zero-byte files are ignored for duplicate grouping.
- The view shows the top duplicate groups by wasted bytes.
Duplicate detection is intentionally conservative: it does not hash file contents. Treat the list as a fast lead generator, then verify files before deleting anything important.
The optional AI Advice pane uses Microsoft Foundry Local to assess every path in the selected duplicate set. File metadata and paths stay on the computer; the app does not upload files or send them to a hosted model.
- Open Settings and enable local AI advice.
- Refresh the model catalog and select a model. The recommended model appears first.
- Click Prepare to download and load the model, then click Run Model Probe.
- After all four deterministic probe scenarios pass, return to Duplicates, select a file occurrence, and click Analyze.
The model returns Keep, Delete candidate, or Review for each supplied path. Results are advisory and never delete files. The app requires exact supplied paths, validates the complete response, and changes deletion advice for protected locations to Review. Because duplicates are matched by name and size rather than content hash, manually verify contents before deleting anything.
Foundry runs in an isolated worker process so model/runtime failures do not take down the main window. The first catalog refresh may install local execution-provider components, and preparing a model may download several gigabytes. Models are not bundled in the installer.
You can right-click folder and file rows to open the normal Windows shell menu for that item.
- Right-click folders in the tree or Details folder grid.
- Right-click files in the Details file grid or Duplicates grid.
- Delete actions refresh the current scan model after Windows removes the file or folder.
- Rename actions update the scan model so paths stay consistent.
If a file or folder no longer exists, the app refreshes the current view from the scan's remaining file list.
Open the Settings tab to adjust logging and Foundry Local behavior.
- Use the theme glyph in the window header to cycle through Auto, Light, and Dark. Auto is the default and follows the Windows app theme.
- Log level controls how verbose the app log is.
- Log every SDK file result records every accepted file path from the Everything SDK.
- Write logs to debug output mirrors log lines to debug listeners.
- Retained log files controls how many historical log files are kept.
- Enable local AI advice controls whether qualified-model analysis is available in the Duplicates tab.
- Model, Prepare, and Run Model Probe manage the local model and its qualification.
- Inference timeout limits each model recommendation request.
Settings are saved to:
%LOCALAPPDATA%\EverythingDiskUsage\settings.json
The Settings tab also shows the settings file path, log folder path, and current log file path.
Check that Everything Search is installed, running, and indexed. Open Everything Search directly and wait for indexing to finish, then scan again.
Everything Disk Usage can only report what Everything has indexed. Confirm that Everything is allowed to index the drive or folder you are scanning.
The app can still install, but it will not be useful until Everything Search is installed and running. The installer can open the Everything download page for you.
The app writes logs to:
%LOCALAPPDATA%\EverythingDiskUsage\logs\
Each run creates a timestamped EverythingDiskUsage-*.log file. Logs include startup/shutdown, UI operations, scan progress, folder/file detail population, Everything SDK query setup, SDK result processing, errors, and cleanup.
Set EVERYTHING_DISK_USAGE_LOG_EACH_FILE=1 before launching the app to log every accepted SDK file result. Without that switch, the scanner logs the first few files and periodic progress samples to avoid generating very large logs on full-drive scans.
cd D:\EverythingDiskUsage
dotnet run --project .\src\EverythingDiskUsage.csprojEverything Search must still be installed, running, and indexed. The app includes Everything64.dll in the repository and copies it beside the executable during build.
Build a Debug binary:
cd D:\EverythingDiskUsage
dotnet build .\src\EverythingDiskUsage.csproj -c DebugBuild a Release binary:
dotnet build .\src\EverythingDiskUsage.csproj -c ReleaseThe normal build output is written under:
src\bin\<Configuration>\net11.0-windows\
Create a self-contained Windows x64 publish output and rebuild the Inno Setup installer:
dotnet publish .\src\EverythingDiskUsage.csproj -c Release -r win-x64 --self-contained -o .\artifacts\publishIf Inno Setup 6 is installed machine-wide, dotnet publish builds the installer automatically at:
installer-output\EverythingDiskUsage-Setup-1.0.0.exe
Copy the rebuilt installer into downloads\ when you want the repository link to point at the newest checked-in installer:
Copy-Item .\installer-output\EverythingDiskUsage-Setup-1.0.0.exe .\downloads\EverythingDiskUsage-Setup-1.0.0.exe -ForceIf ISCC.exe is installed somewhere non-standard, pass its path explicitly:
dotnet publish .\src\EverythingDiskUsage.csproj -c Release -r win-x64 --self-contained -o .\artifacts\publish /p:IsccPath="C:\Path\To\ISCC.exe"Build the installer without committing or publishing:
.\scripts\build-all-installers.ps1 -Version 1.0.0Preview the complete release plan without changing files:
.\scripts\build-all-installers.ps1 -Push -ReleaseMode Published -WhatIfBuild the next patch version, commit all pending repository changes, push the current branch, and publish the GitHub release:
.\scripts\build-all-installers.ps1 -Push -ReleaseMode PublishedOmit -ReleaseMode Published to choose interactively between a draft and a
published release. The script requires GitHub CLI (gh) for release creation.
Run the automated test suite with:
dotnet test .\EverythingDiskUsage.slnx --configuration DebugRun tests with coverage collection:
dotnet test .\EverythingDiskUsage.slnx --configuration Debug --collect:"XPlat Code Coverage" --results-directory .\TestResultsThe Azure publishing script builds the app, creates the release ZIP, uploads it to the private downloads container, and updates the Everything Disk Usage card on the static download site:
.\scripts\publish-to-azure.ps1 -Configuration ReleaseThis requires Azure CLI access to the installmonitordl storage account in the rg-installmonitor-download resource group.
Everything Disk Usage is released under the 0BSD license. See LICENSE.