Worktrunk is an open-source command-line tool that simplifies Git worktree management, especially for developers running multiple AI coding resources at the same time. Built by Max Sixty, it is a convenient interface for creating, modifying, registering, joining, and deleting active worktrees without having to use lengthy Git commands.

Git work trees allow multiple branches from the same repository to have separate working directories. This is especially useful when agents like Claude Code and OpenAI Codex are run in parallel, as each agent can work independently without affecting the other agent’s files.

Worktrunk builds on this functionality with additional automation, collaborative navigation, and development workflow features.

Features

Simplified Git Worktree Management

Worktrunk provides three core commands for everyday worktree operations:

  • wt switch switches between existing worktrees or creates new ones.

  • wt list displays worktrees with branch status and change information.

  • wt remove removes worktrees and their associated branches.

For example, creating a new feature branch and worktree requires only:

wt switch --create feature-auth

Worktrunk automatically creates the branch, prepares its working directory, and switches the shell into it.

This eliminates much of the repetitive work involved in managing Git worktrees manually.

Parallel AI Coding Agents

Worktrunk is particularly useful for developers working with multiple AI coding agents.

It can create a worktree and immediately launch an agent using the -x option:

wt switch -x claude --create feature-a -- "Implement user authentication"

Developers can repeat this process for different tasks, giving each agent its own branch and working directory.

This approach works with different command-line coding agents, including Claude Code and Codex. Worktrunk does not provide its own AI model or require users to replace their existing coding tools.

Automated Development Hooks

Worktrunk includes configurable hooks that execute commands during different stages of the worktree lifecycle.

Examples include:

  • Pre-start hooks for installing dependencies.

  • Post-start hooks for preparing development environments.

  • Pre-merge hooks for running tests and validation.

  • Post-merge hooks for cleaning up temporary resources.

Hooks can automate repetitive setup tasks whenever a new worktree is created.

For projects using Node.js, developers can configure dependency installation automatically. Other workflows can prepare databases, environment files, development servers, or additional resources.

Interactive Worktree Picker

Worktrunk provides an interactive interface for navigating existing worktrees.

Instead of remembering branch names and directory paths, developers can browse available worktrees and inspect their status.

The picker can display additional information, including Git diffs, commit history, pull requests, comments, and CI status.

This becomes increasingly useful when managing many parallel development tasks.

Shared Build Caches

One challenge with Git worktrees is that each working directory can require its own dependency installation and build output.

Worktrunk addresses this through copy-on-write build cache sharing.

It supports sharing directories such as node_modules/ and Rust's target/ across worktrees on supported filesystems, including APFS, Btrfs, and XFS.

This can significantly reduce redundant disk usage and avoid repeatedly rebuilding dependencies.

The feature depends on filesystem support and appropriate configuration. It should not be assumed to work identically on every operating system or storage configuration.

Git Merge and Cleanup

Worktrunk provides a simplified merge workflow that can handle committing, rebasing, merging, and cleaning up a completed worktree.

For example:

wt merge main

The command can prepare changes, generate a commit message when configured, merge the branch into the target, and remove the completed worktree.

This reduces the number of manual Git operations required to finish a development task.

Pull Request Integration

Worktrunk supports switching directly to pull request branches.

For example:

wt switch pr:123

It can also display CI status and pull request information through its listing and interactive navigation features.

This makes it easier to move between local development branches and remotely hosted contributions.

Custom Commands and Configuration

Worktrunk supports aliases, custom commands, configurable worktree paths, and per-branch variables.

Developers can use these features to build project-specific workflows without maintaining separate shell scripts for every operation.

Its configuration system also supports LLM-generated commit messages, allowing developers to generate commit descriptions from their changes when an appropriate AI tool is configured.

Download Worktrunk v0.81.0 - Software Mirrors

Worktrunk v0.81.0 for Windows

worktrunk-x86_64-pc-windows-msvc.zip | 16.65 MB

Worktrunk v0.81.0 for macOS

worktrunk-x86_64-apple-darwin.tar.xz | 6.87 MB

worktrunk-aarch64-apple-darwin.tar.xz | 6.27 MB

Worktrunk v0.81.0 for Linux

worktrunk-x86_64-unknown-linux-musl.tar.xz | 7.99 MB

worktrunk-aarch64-unknown-linux-musl.tar.xz | 7.01 MB

Others Download related to Worktrunk v0.81.0

wt.rb | 2.58 kB

worktrunk-installer.sh | 52.25 kB

worktrunk-installer.ps1 | 20.86 kB

source.tar.gz | 3.91 MB

sha256.sum | 610 B

Worktrunk v0.81.0 Source Code

Worktrunk v0.81.0 Source code (zip)

Worktrunk v0.81.0 Source code (tar.gz)

Worktrunk v0.81.0 Release Notes:

Release Notes

Improved

  • Hooks can prompt: hooks that run serially in the foreground (pre-* hooks, wt hook --foreground) inherit wt's stdin, so gum confirm 'trust this worktree?' && mise trust works as a hook. Concurrent and background hooks read EOF. (Breaking: hooks no longer receive their context as JSON on stdin; pass values as template arguments, or whole objects with the new tojson filter. wt step for-each keeps its JSON.) Docs (#3129, #4451, #4524, thanks @endigma for #3093)

  • Experimental wt switch --path: places one worktree outside the worktree-path template and keeps its branch name, as in wt switch --create feature/JIRA-1234 --path ../dark-mode. Commands reach it by branch or by path. Docs (#4317, thanks @christowiz for #1982)

  • wt merge --branch: merges another worktree's branch without leaving the current worktree. Commit, squash, rebase, and hooks run in the source worktree, and cleanup removes it. Docs (#4360, thanks @liuyinchu)

  • OpenCode 2 worktrees go through Worktrunk: after re-running wt config plugins opencode install, worktrees OpenCode creates and removes use wt switch --create and wt remove, as the Claude Code plugin's do. OpenCode's worktree.directory stops applying, and new branches start from the default branch. Like the Claude Code plugin, it opens an existing branch's worktree, which OpenCode's remove then removes. Docs (#4489, thanks @rzane for #4488)

  • Faster bulk removal: removal hands files to one background deleter per repository, not one rm -rf each, and deletes read-only directories it used to leave in .git/wt/trash. wt step prune no longer stalls when two worktrees share a directory name, as Codex worktrees do. (Breaking: prune runs pre-remove hooks concurrently; hooks writing shared state must coordinate.) (#4516, #4361, #4337)

  • wt switch - returns to the worktree it left, including a detached worktree or the second checkout of a branch, and wt remove - targets the same worktree. Pre-switch hooks see the resolved destination, so wt switch origin/feature passes branch=feature. (Breaking: from a bare repository's directory, commands given no argument act on the worktree of the branch HEAD names.) (#4497, #4496, #4521)

  • Template errors name the variable: {{ brnch }} reports undefined variable brnch and suggests branch. (Breaking: filters such as sanitize reject an undefined input, so {{ upstream | sanitize }} needs a guard or default on a branch with no upstream.) (#4350, #4440)

  • Templates use MiniJinja 3 (Breaking): objects and lists render Python-style ({'port': 3000}) instead of as JSON, so use | tojson where JSON is needed; round rounds half to even; title capitalizes fewer word boundaries (feat/x-y becomes Feat/x-Y); division by zero errors. (#4442, #4524)

  • CI status shows ⚠ for any failed forge call, where some failures, such as a Gitea 404, used to leave the cell blank. A forge CLI with no login for the host still leaves it blank, and wt config show --full names the host it checked. (#4436, #4449)

  • Removal explains itself: --force lists the changes it discards, and a refusal over untracked files suggests git stash -u. A multi-target wt remove continues past a target whose pre-remove hook or removal fails, and exits 1. A branch-only removal names the detached worktree left at the branch's path. (#4347, #4357, #3791, #4313, thanks @chachi for #3769)

  • Squash backups print a recovery command: squashing staged changes prints the refs/wt-backup/ ref and a wt switch --create --base command that restores them in a new worktree, even after wt merge removed the original. (#4347)

  • Shell integration: wt merge and wt remove suggest restarting the shell when integration is installed but inactive. On Unix, the PowerShell profile comes from the installed pwsh, so a custom XDG_CONFIG_HOME works. (Breaking: installing for Nushell requires nu, and for PowerShell on Unix, pwsh.) (#4343, #4422, #4436)

  • System config has one path per platform: /Library/Application Support/worktrunk/config.toml on macOS, %PROGRAMDATA%\worktrunk\config.toml on Windows, /etc/xdg/worktrunk/config.toml elsewhere, or WORKTRUNK_SYSTEM_CONFIG_PATH. (Breaking: $XDG_CONFIG_DIRS is ignored and macOS no longer falls back to /etc/xdg; a file there stops loading without a warning.) (#4169)

  • expected_worktree_path_of_branch() template function: returns where wt switch --create would put a branch's worktree, even after the worktree at that path has moved to another branch. (#4502, thanks @BenJurewicz for #4480)

  • Smaller improvements: wt switch --create reports an invalid branch name in one line instead of Git's error; hook failures name the hook's source (pre-merge user:lint failed (exit code 7)); the wt list loading footer says how many rows fit; wt step for-each labels detached worktrees by path; install-statusline keeps settings.json's key order; the Codex plugin shows the Worktrunk icon. (#4349, #4352, #4333, #4498, #4420, #4327)

Fixed

  • Foreground hooks that change terminal settings: a hook that touched the terminal, such as turbo on a cache miss, was stopped by SIGTTOU and wt hung. Hooks now run in the shell's job group. (Breaking: Nushell integration requires Nushell 0.113+; on Ctrl-C wt dies by the signal instead of exiting 130, so a calling script stops too.) (#4400, #3129, thanks @user753 for #4389 and @willparsons for the turbo report)

  • Formatter pre-commit hooks during a squash: in wt merge and wt step squash, a hook's edits missed the squash commit, and the dirty worktree blocked removal. All commit paths now stage, run the hooks, and stage again. (Breaking: wt step commit and wt merge --no-squash stage first too, so index-reading hooks like pre-commit run can block them.) (#4321, #4520)

  • wt step copy-ignored on Linux: copied files took the copy's time, so cargo rebuilt much of a copied target/; they now keep the source's mtime, as they already did on macOS and Windows. A reflink rejected with EPERM now falls back to a copy. (#4319, #4518, #4434, thanks @dmy-gh for #4248, @paul-hansen for the repro, and @HN05 for #4433)

  • wt list status symbols: JSON symbols strings use the table's order (⊘_, was _⊘), and the flag for an off-template path or duplicate checkout is a dim ⚐ instead of ⚑. (Breaking: consumers matching these strings need updating.) (#4335, #4334)

  • wt switch: --create branches from a remote-tracking default branch such as git-p4's p4/master instead of HEAD; switching by path, as the picker does, gives pre-switch hooks the destination's variables; the bare-repo warning names the configured path. (#4482, #4493, #4491, thanks @cazador481 for #4481 and @dlecan for #4492)

  • Agent plugins: claude --worktree NAME opens the worktree of an existing branch NAME rather than failing, so a kept session can be reopened. A clean exit then runs wt remove on it, which deletes its ignored files and, if integrated, the branch. The Codex plugin's Windows hooks parse under PowerShell. (#4487, #4241, thanks @edziegle for #4486 and @wstczyw for #4239)

  • wt switch picker: a per-project pager applies to the diff preview; a selected branch-only row no longer draws light text on the light selection bar; an orphan branch's log tab shows its commits. (#4398, #4408, #4470, thanks @willparsons for reporting the contrast bug)

  • wt step relocate: a swap moves the shell with its worktree; relocating the main worktree leaves a shell in a nested worktree where it is; branches whose sanitized names collide no longer relocate partway. (#4325, #4256, thanks @Duang777)

  • Aliases: --args=X and --vars=X reach {{ args }} instead of being dropped, and completion after an alias no longer offers --dry-run or --var. (#4391, #4329)

  • Config errors: a malformed approvals.toml no longer aborts hooks, aliases, or commits run with --yes; one invalid WORKTRUNK_* override no longer discards the valid ones. (#4467, #4348)

  • Windows: an Azure CLI installed through WinGet (az.cmd) is found, and Git Bash is found through the first git on PATH, including custom installs where it resolves under Git\mingw64\bin. (Breaking: without git on PATH, Git Bash is no longer found at its default install locations.) (#4402, #4411, #4452, thanks @wmmc88 for #4401 and @LeoWSY-hashblue for #4410)

  • Forge fixes: Azure DevOps CI no longer shows a same-named branch's PR from another repository in the project, and pr:N suggests a dev.azure.com URL for SSH remotes; Gitea fork pushes keep the server's clone URL; tea logins in its default macOS and Windows locations are found; gh and glab protocol settings apply per host. (#4326, #4364, #4495, #4420)

  • Fixes for unusual setups: valueless git config keys such as core.fsmonitor read as true; untracked files named like :x no longer break wt list's diff; worktree paths containing newlines or non-UTF-8 bytes; the statusline's branch label with custom worktree-path templates; zsh compinit advice names the right file; IPv6 ssh:// remotes get distinct approvals and projects] settings (Breaking: re-save them). ([#4314, #4320, #4453, #4495, #4421, #4526)

Documentation

  • wt config shell --help is the shell setup reference, covering init, install, uninstall, and diagnostics. (#4344)

  • Doc corrections: the wt config state vars clear example, wt list JSON null causes, every supported forge in CI help, --path worktrees in wt list, the bare-repo navigation example, and Alt as Option on macOS. (#4363, #4359, #4439, #4328, #4483, #4412, thanks @wmcnamee-coreweave for #4474)

  • Recipes: the bare repository layout sets origin's fetch refspec; the aichat example passes --code to drop a reasoning model's <think> block; the OpenCode command uses Claude Haiku 5.5; wt-tui joins the custom subcommand examples. (#4323, #4332, #4432, #4386, thanks @gitNetw0rk for #4322, @FenjuFu, and @marcusjihansson)

  • The wt-switch-create skill reads the new repository's root CLAUDE.md and AGENTS.md when it moves a session into another repository's worktree. (#4414)

Internal

  • Library API changes (Breaking library API): Repository::project_identifier returns a ProjectId, which config accessors take in place of &str; Repository::require_worktree also returns the branch; Repository::run_command takes AsRef arguments; UserConfig::expected_worktree_path is new. (#4526, #4496, #4453, #4502)

  • MSRV is Rust 1.98. (#4367)

  • A hidden wt config plugins claude hook subcommand handles every hook event the Claude Code plugin uses, toward listing the plugin in Anthropic's plugin directory. (#4304)

  • Release signing and crates.io OIDC run in the tag-only release environment. (#4415)

Install worktrunk 0.81.0

Install prebuilt binaries via shell script

curl --proto '=https' --tlsv1.2 -LsSf https://github.com/max-sixty/worktrunk/releases/download/v0.81.0/worktrunk-installer.sh | sh && wt config shell install

Install prebuilt binaries via powershell script

powershell -ExecutionPolicy Bypass -c "irm https://github.com/max-sixty/worktrunk/releases/download/v0.81.0/worktrunk-installer.ps1 | iex"; git-wt config shell install

Install prebuilt binaries via Homebrew

brew install worktrunk && wt config shell install

Download worktrunk 0.81.0

Install via Cargo

cargo install worktrunk && wt config shell install

Install via Winget (Windows)

winget install max-sixty.worktrunk && git-wt config shell install

Install via AUR (Arch Linux)

paru worktrunk-bin && wt config shell install


Performance and Compatibility

Worktrunk is written in Rust and distributed as a native command-line application. It does not require a background server or introduce an additional AI runtime.

Its main workload involves Git operations, filesystem management, shell integration, and optional workflow automation. Consequently, its performance depends more on repository size, filesystem behavior, and configured hooks than on AI model processing.

It supports macOS, Linux, and Windows. Installation is available through Homebrew, Cargo, Winget, and several Linux package managers.

Shell integration is available for supported shells, allowing commands such as wt switch to change the current directory directly.

Windows users should note that wt can conflict with the Windows Terminal command alias. The Winget installation provides git-wt as an alternative.

System Requirements

Worktrunk requires:

  • Git installed and accessible through the system PATH.

  • A supported operating system: Windows, macOS, or Linux.

  • A compatible shell for directory-switching integration.

  • A Git repository for worktree operations.

Optional requirements depend on the features being used:

  • Cargo and Rust for building from source.

  • jq for certain integrations and scripts.

  • A compatible AI coding agent for automated agent workflows.

  • A supported filesystem for copy-on-write cache sharing.

Pros and Cons

Pros

  • Simplifies Git worktree creation and management.

  • Designed for parallel AI coding workflows.

  • Supports Claude Code and other command-line agents.

  • Automates development tasks through hooks.

  • Provides interactive worktree navigation.

  • Supports shared build caches.

  • Includes merge, cleanup, and pull request workflows.

  • Written in Rust and open-source.

  • Supports Windows, macOS, and Linux.

Cons

  • Requires familiarity with Git worktrees to use its advanced features effectively.

  • Shell integration requires additional configuration.

  • Copy-on-write cache sharing depends on filesystem compatibility.

  • Some advanced features require external tools.

  • Does not provide AI models or coding agents itself.

How to Install

The easiest installation method on macOS and Linux is Homebrew:

brew install worktrunk
wt config shell install

For installation through Cargo:

cargo install worktrunk
wt config shell install

On Windows, use Winget:

winget install max-sixty.worktrunk
git-wt config shell install

Arch Linux users can install it through Pacman:

sudo pacman -S worktrunk
wt config shell install

After installation, restart the terminal or reload the shell configuration.

Verify the installation with:

wt --help
wt list

To create a new worktree:

wt switch --create feature-test

The official documentation provides additional configuration examples for hooks, aliases, build caches, and parallel AI agent workflows.

Final Verdict

Worktrunk is an excellent utility for developers who frequently work with multiple Git branches, especially those using AI coding agents.

Its biggest advantage is that it makes Git worktrees considerably easier to manage. Features such as automated hooks, shared build caches, interactive navigation, and integrated merge workflows turn what would otherwise be repetitive Git operations into a more organized development process.

It is not an AI coding assistant itself, but it complements tools such as Claude Code and Codex by providing isolated environments for parallel development.

For developers who regularly run multiple coding agents or maintain several features simultaneously, Worktrunk is a practical addition to their development toolkit.

Sources


Worktrunk v0.81.0
Free
Download

Software Informations:
Developer:

Operating System:
Windows / macOS / Linux
Date Added:
2026-11-03T14:37:02.000Z

DiskRaptor is a modern open source disk space analyzer designed to help you quickly understand what is consuming storage on your computer. Built with Rust and Tauri 2, it is positioned as a faster, more modern alternative to tools such as WinDirStat and DaisyDisk.

What makes DiskRaptor particularly interesting is its focus on handling very large file systems. The project uses native Rust scanning components and a virtual tree interface designed to remain responsive even when working with millions of files.

What Is DiskRaptor?

DiskRaptor is a cross platform disk usage analyzer for Windows, macOS, and Linux.

Its main purpose is simple. You select a drive or directory, let DiskRaptor scan it, and then explore which folders and files are consuming the most space.

The project is built with Rust and Tauri 2. The native scanning engine handles file system traversal while the interface runs inside the operating system's native webview. This keeps the application relatively lightweight compared with applications that bundle an entire Chromium runtime.

Fast Disk Scanning

Performance is one of DiskRaptor's main selling points.

The project uses a parallel scanning engine on macOS and a reliable walkdir-based engine on Windows and Linux. It is designed to handle drives containing millions of files, with the project documenting support for scans involving up to 20 million nodes.

Access denied folders are handled gracefully as well. Instead of completely stopping a scan because it encounters a protected directory, DiskRaptor logs the problem and continues scanning.

For users with large SSDs, hard drives, external drives, or development machines containing huge numbers of files, this approach can make a significant difference.

Virtual Tree View

DiskRaptor uses a virtual tree interface that renders only the items currently visible on screen.

This is important when working with extremely large directory structures. Loading millions of files into a conventional browser interface would quickly become impractical.

The project claims that its virtual tree can handle more than 10 million files while remaining responsive.

You can navigate through folders, inspect file sizes, and identify which directories are responsible for consuming storage.

Multiple Visualization Options

DiskRaptor doesn't rely on a single visualization.

It provides several ways to understand disk usage, including:

  • Pie charts

  • Squarified treemaps

  • Bar charts

  • Galaxy 3D visualization

The application also provides a Top 50 list of the largest files. Selecting an item in the visualization can take you directly to the corresponding location in the directory tree.

The treemap is particularly useful when you want a quick visual overview of a large drive.

Find Large Files

Finding large files is one of the most practical reasons to use a disk analyzer.

DiskRaptor highlights the largest files and allows you to inspect their location directly.

This can quickly reveal things such as old installers, virtual machine images, Docker data, video files, build artifacts, backups, and other files that have quietly consumed hundreds of gigabytes.

Duplicate File Detection

DiskRaptor also includes duplicate detection.

It uses content hashing with xxHash3 to identify files that contain the same data.

This is more useful than simply comparing filenames because files with different names can still contain identical content.

Duplicate detection can be particularly helpful for large collections of photographs, videos, archives, downloads, and project backups.

Empty Folder Finder

The application can identify empty directories and allow you to jump directly to them.

This is a relatively small feature, but it can be useful when cleaning up development directories, extracted archives, or old project structures.

File Search and Filtering

DiskRaptor provides search functionality for scanned results.

You can search by filename pattern and filter results by file type, including:

  • Images

  • Video

  • Audio

  • Archives

  • PDF files

This makes it easier to locate specific categories of files without manually navigating the entire directory tree.

File Management

DiskRaptor isn't purely a visualization tool.

Its context menus provide actions such as opening a file in Explorer or Finder, opening a terminal, viewing properties, copying a path, moving an item to the trash, and scanning a particular folder.

The ability to move files to the trash directly from the application makes cleanup considerably more convenient.

Because file deletion is potentially destructive, users should still verify what they are removing before performing cleanup operations.

Trash Recovery

Another useful feature is trash management.

DiskRaptor can browse items in the operating system's trash, restore them, or permanently delete them.

This gives the application some basic recovery functionality rather than treating deletion as a one way operation.

Reports and Export

DiskRaptor can export disk analysis results in several formats.

Supported options include:

  • HTML reports

  • CSV

  • JSON

The HTML report can be shared as a standalone report containing statistics and charts. CSV and JSON exports are useful when you want to process the results with another application or script.

This is particularly useful for administrators who need to document storage usage rather than simply inspect it interactively.

Cross Platform Support

DiskRaptor supports:

  • Windows

  • macOS

  • Linux

The project provides a Windows NSIS installer, a universal macOS binary supporting both Intel and Apple Silicon, and a DEB package for Debian and Ubuntu.

This makes it considerably more versatile than Windows-only disk analyzers.

Interface and Customization

The interface supports both dark and light modes, with the selected setting saved for future sessions.

It also supports 25 languages with automatic language detection and a language switcher.

Drag and drop is supported as well, allowing folders to be dragged from Explorer or Finder into DiskRaptor for scanning.

The interface also has resizable panels, keyboard navigation, and contextual menus.

Technology

DiskRaptor is built around Rust and Tauri 2.

The architecture separates the native layer from the interface. Rust handles scanning and native operations while JavaScript handles the user interface through the operating system's webview.

That means DiskRaptor does not need to ship its own Chromium browser engine.

On Windows it uses WebView2, macOS uses WKWebView, and Linux uses WebKitGTK.

For a disk utility that needs to remain responsive while processing very large directory trees, this is a sensible architecture.

Download DiskRaptor v1.0.51 - Software Mirrors

DiskRaptor v1.0.51 for Windows

DiskRaptor.exe | 14.26 MB

DiskRaptor-windows-x64.msix | 5.34 MB

DiskRaptor-1.0.51-windows-x64.msi | 8.05 MB

DiskRaptor-1.0.51-windows-x64.exe | 5.97 MB

DiskRaptor-1.0.51-windows-x64-silent.msi | 5.36 MB

DiskRaptor-1.0.51-windows-x64-silent.exe | 3.84 MB

DiskRaptor v1.0.51 for macOS

DiskRaptor-1.0.51-macos-universal.pkg | 13.87 MB

DiskRaptor-1.0.51-macos-universal.dmg | 14.76 MB

DiskRaptor v1.0.51 for Linux

DiskRaptor-1.0.51-linux-x86_64.rpm | 7.53 MB

DiskRaptor-1.0.51-linux-x86_64.deb | 7.53 MB

DiskRaptor-1.0.51-linux-x86_64.bin | 11.27 MB

DiskRaptor-1.0.51-linux-x86_64.AppImage | 82.37 MB

DiskRaptor v1.0.51 Source Code

DiskRaptor v1.0.51 Source code (zip)

DiskRaptor v1.0.51 Source code (tar.gz)

DiskRaptor v1.0.51 Release Notes:

Full Changelog: v1.0.50...v1.0.51

Pros

  • Open source

  • Free to use

  • Windows, macOS, and Linux support

  • Built with Rust

  • Tauri 2 architecture

  • Fast directory scanning

  • Designed for millions of files

  • Virtual tree view

  • Treemap visualization

  • Pie charts

  • Bar charts

  • Galaxy 3D view

  • Top 50 largest files

  • Duplicate file detection

  • Empty folder detection

  • Filename search

  • File type filtering

  • Trash recovery

  • Direct file management

  • HTML report export

  • CSV export

  • JSON export

  • Dark and light modes

  • 25 languages

  • Drag and drop support

  • Apple Silicon support

Cons

  • The project is still relatively young

  • Some advanced features may need further refinement

  • Scanning very large drives can still consume substantial CPU and storage I/O

  • Deleting files directly from a disk analyzer requires caution

  • The visualization features may be more than users need for simple disk cleanup

  • Linux packaging currently focuses on DEB distributions

DiskRaptor vs WinDirStat

WinDirStat remains one of the best known Windows disk usage analyzers, but DiskRaptor takes a more modern approach.

DiskRaptor is cross platform, built with Rust and Tauri, supports multiple visualization modes, includes duplicate detection and trash recovery, and is designed around very large directory trees.

WinDirStat has the advantage of being a mature and well established Windows application.

For users who want a modern cross platform alternative, DiskRaptor is much more interesting.

DiskRaptor vs DaisyDisk

DaisyDisk is known for its polished macOS interface and strong visualization capabilities.

DiskRaptor brings a similar visual approach to Windows and Linux while also supporting macOS.

Its open source nature is another major difference.

For someone who wants a free and open source disk analyzer that works across multiple operating systems, DiskRaptor is a compelling alternative.

Who Should Use DiskRaptor?

DiskRaptor is a good choice for anyone who regularly needs to answer the question, "What is using all my disk space?"

It is especially useful for:

  • Developers

  • System administrators

  • Power users

  • NAS and storage users

  • Large media collections

  • Users managing external drives

  • People cleaning old project files

  • Anyone with a rapidly filling SSD

The large file and duplicate detection features are particularly useful for development machines where build artifacts, package caches, Docker files, virtual machines, and old repositories can accumulate surprisingly quickly.

Final Verdict

DiskRaptor is an impressive open source disk space analyzer with a clear focus on performance and usability.

Its Rust based scanning engine, virtual tree interface, multiple visualizations, duplicate detection, file filtering, and report export make it much more than a basic folder size viewer.

The cross platform support is another major advantage. You can use the same general workflow on Windows, macOS, and Linux rather than maintaining different disk analysis tools for each operating system.

The project is also under active development. The GitHub repository currently shows more than 800 commits, six open pull requests, and recent updates in August 2026.

It is still less mature than long established applications such as WinDirStat, so some users may encounter rough edges. Nevertheless, the architecture and feature set make DiskRaptor one of the more interesting modern alternatives in this category.

If you want a fast, open source, cross platform disk space analyzer, DiskRaptor is well worth trying.

DiskRaptor v1.0.51
Free
Download

Software Informations:
Developer:

Operating System:
Windows / Linux / macOS
Date Added:
2026-11-03T14:07:17.204Z

GDevelop is an open source, cross-platform game engine that makes game development accessible without requiring traditional programming skills. It supports 2D, 3D, and multiplayer games and can export projects to the web, desktop, Android, iOS, Steam, and other platforms. Its event-based visual programming system, ready-made behaviors, extensive asset library, and optional AI assistance make it one of the most approachable game development tools available today.

GDevelop takes a different approach from traditional engines such as Unity or Unreal Engine. Instead of requiring developers to write large amounts of code, its event system lets users construct game logic using conditions and actions.

For example, a game rule can be expressed as a simple sequence of "If" and "Then" events. This makes the engine particularly approachable for beginners, educators, artists, and designers who want to create games without first learning a programming language.

At the same time, GDevelop is not limited to simple projects. JavaScript can be added when more advanced functionality is needed, while extensions and behaviors allow developers to expand the engine considerably.

Key Features of GDevelop

No-Code Game Development

The event system is the heart of GDevelop. Users can create gameplay mechanics visually by combining conditions and actions rather than writing traditional code.

This makes common tasks such as player movement, collisions, enemy behavior, scoring, menus, and object interactions considerably easier for beginners.

2D and 3D Game Development

GDevelop has expanded significantly beyond its original 2D focus. The current engine supports both 2D and 3D game development, with 3D rendering powered by Three.js alongside PixiJS for 2D graphics.

The newer 3D workflow includes an editor where developers can place objects and work with their game world more directly, making 3D development much more accessible than it was in earlier versions.

Ready-Made Behaviors

GDevelop includes a large collection of behaviors and extensions that provide common game functionality without requiring developers to build everything from scratch.

These can handle things such as platformer movement, physics, pathfinding, cameras, tweens, and other gameplay systems. The official website currently advertises more than 130 ready-made behaviors and extensions.

JavaScript Support

Although GDevelop is designed around no-code development, experienced developers are not restricted to the visual event system.

JavaScript can be used to extend gameplay logic, interact with APIs, access external data, and create functionality that is difficult to express entirely through events.

Multiplayer

GDevelop includes multiplayer capabilities and online services for games that require features such as player accounts, lobbies, and leaderboards.

This gives developers the ability to build more connected games without having to construct every backend service from scratch.

AI-Assisted Development

AI assistance has become part of the GDevelop workflow. The AI agent can help users create game logic, answer development questions, and accelerate prototyping.

The important part is that AI remains optional. Users can continue creating games entirely through the traditional GDevelop editor and event system.

Asset Store

GDevelop includes access to a large collection of ready-made assets and templates. The official platform currently lists more than 2,000 assets, including free resources and premium packs.

This is particularly useful for beginners who want to build a playable prototype without creating every visual asset themselves.

Cross-Platform Export

Games can be exported to multiple platforms, including:

  • Windows

  • macOS

  • Linux

  • Android

  • iOS

  • HTML5

  • Steam

  • Epic Games Store

  • Microsoft Store

This makes GDevelop useful for developers who want to release the same project across several platforms without maintaining completely separate codebases.

Download GDevelop v5.6.286 - Software Mirrors

GDevelop v5.6.286 for Windows

GDevelop-5-Setup-5.6.286.exe | 147.92 MB

GDevelop-5-5.6.286-win.zip | 195.55 MB

GDevelop v5.6.286 for macOS

GDevelop-5-5.6.286-universal.dmg | 269.78 MB

GDevelop-5-5.6.286-universal-mac.zip | 260.05 MB

GDevelop v5.6.286 for Linux

GDevelop-5-5.6.286.AppImage | 201.61 MB

GDevelop-5-5.6.286-linux.zip | 192.04 MB

GDevelop-5-5.6.286-arm64.AppImage | 201.91 MB

GDevelop v5.6.286 Source Code

GDevelop v5.6.286 Source code (zip)

GDevelop v5.6.286 Source code (tar.gz)

GDevelop v5.6.286 Release Notes:

📂 Folders now available in the Project Manager

Use folders to organize scenes, external layouts, external events and gameplay tests in the project manager. This is helpful for large games, for example when you have a lot of external layouts for multiple levels in a game. !Folders used in the project manager

🤖 Connect GDevelop to your local AI agent ("MCP")

An experimental "MCP" connection is now available to plug your own agent to GDevelop's tools. This is useful if you want to experiment building, debugging or learning with an AI agent like Claude Code, Codex, etc... !Claude Code connected to GDevelop Read more about how to setup a local agent in the documentation. This MCP, and the integrated GDevelop AI agent, will be improved if they are proving useful.

🌊 New ready-to-use 3D water and liquids

When adding a new 3D water object to your scene, you can now select a ready-made version from the GDevelop asset store: !A scene with 3D water and the dialog showing some ready-to-use, pre-made 3D water assets They work well with the 3D terrains or, more generally, any 3D game.

💝 Improvements

  • Make gameplay tests more reliable and support multiplayer games
  • Add a "synchronized over the network" setting to behaviors:
  • When disabled, behaviors are not synchronized, which is useful for behaviors like "Input Mappers".
  • Add a shadow normal bias to directional lights to avoid shadow acne.
  • Add Bluesky to the community links of the profile
  • Redesign the order of the properties of 3D boxes in the properties panel.
  • For custom objects and custom behaviors, properties are shown grouped per folder, and the groups are following the same order as these folders.
  • Add pre-made effects that can be added to objects or 2d layers (tints & status colors)
  • Add conversions across the boundary of custom objects ("ToParent"/"FromParent" expressions)
  • Also add 3D versions of distances conditions/expressions and "turned toward" condition.
  • Use the discrete graphics card and warn when the 3D rendering is slow (useful for devices with multiple GPUs or with hardware acceleration disabled)
  • Load big projects faster and use less memory to load and save them.
  • Show 3D model previews as icons of 3D objects and model resources.

🐛 Bug fixes

  • Prevent crash from stale multiplayer heartbeat answers (thanks @dajiaohuang!)
  • Fix crash of the 3D editor when adding an object with 3D physics from the asset store
  • Fix instances placed in a scene being deleted on players when the scene is slow to load (multiplayer)
  • Fix some memory leaks in the editor
  • Fix object changes reverted in the in-game editor after switching editors
  • Fix 3D layers wrongly not antialiased when they have post-processing effects
  • Fix some crashes when an Array variable get holes
  • Fix default value for 'block-on-diagnostic-errors' when not specified in the CLI (Thanks @ViktorVovk!)
  • Fix stencil buffer not available on 3D layers with post-processing effects.
  • Fix some effects display, especially CRT
  • Fix crash of the user chip when the profile has no username nor email.
  • Fix othographic camera scale according to the world scale

⚙️ Extensions

  • Set the "mapper" behaviors as not synchronized over the network by default: they translate the inputs of the local player: on the copies of objects owned by other players. They must stay deactivated instead of being re-activated by the owner's game.
  • [Reviewed][Mouse pointer lock] Fix touch look on mobile and screen-ratio-dependent mouse speed (v0.3.2)
  • [Reviewed] Move 3D terrain and 3D water to reviewed extensions
  • [Foliage swaying 3D] Improve performance and fix PBR consistency (v1.3.1)
  • [Marching Squares] Fix drawing only the screen on a 3D layer (v0.6.2)
  • [Pinching gesture] Fix converting positions to the scene on a 3D layer (v1.2.2)

🎨 Assets

  • Add 6 new assets packs from Elv:
* Dark Castle Tileset 16x16 Pixel Art * Fantasy Dreamland Enchanted Forest * Fantasy Dreamland Dungeons 2 * Fantasy Dreamland Dreams * Fantasy Dreamland Castle * Fantasy Dreamland Blacksmith Workshop

🕹 Premium Game Templates

* Replaced the original terrain and ocean models with Terrain3D and Water3D. * Improved grass, vegetation and rock placement using TerrainGrass and ScatteredModels. * Added NavMesh pathfinding and improved lighting, shadows and visual effects. * Optimized models, textures and draw distances for better performance and removed unused resources. * Replaced the original map and lake models with Terrain3D and Water3D. * Optimized vegetation with TerrainGrass and ScatteredModels, replacing 30,000 JavaScript-spawned grass objects. * Improved animal AI with NavMesh pathfinding and added lighting, fog and tone mapping. * Optimized shaders, character models and scene performance, updated the resolution to 1920 × 1080, and fixed player animation speeds.
  • Fixed a bug where using multi transitions in the same scene causes the
extension to create huge amount of shape painter instances.
  • Fixed an issue where shape painter object instances stays in the scene
after the transition has finished.

Performance and User Experience

GDevelop is one of the easier game engines to approach for someone without previous programming experience. The visual event system is intuitive, while the editor provides enough depth to support considerably more sophisticated projects.

The interface also benefits from ready-made behaviors and extensions. Instead of spending hours implementing basic systems, developers can attach existing functionality to game objects and focus more heavily on gameplay design.

Performance is another area where GDevelop has matured. Its engine uses technologies such as PixiJS and Three.js for rendering, while physics extensions include Box2D and Jolt Physics.

There are still limitations compared with highly specialized engines. Developers building extremely complex 3D games may eventually encounter areas where Unity, Unreal Engine, or Godot provide greater depth and control. However, that is not really GDevelop's primary goal.

Its strength is the balance between accessibility and capability.

Pros

  • Free and open source.

  • Excellent for beginners.

  • Supports 2D and 3D development.

  • Visual event-based programming.

  • JavaScript support for advanced users.

  • Large collection of behaviors and extensions.

  • AI-assisted development.

  • Multiplayer capabilities.

  • Large asset and template ecosystem.

  • Cross-platform export.

  • Can be used locally or through cloud-based workflows.

  • Games created with the engine can be distributed commercially without being forced to open source the game.

Cons

  • Advanced 3D development is not as mature as leading professional engines.

  • Visual event systems can become complicated in large projects.

  • Some cloud, publishing, and AI features are tied to paid plans.

  • Developers who prefer conventional programming may find the event system restrictive.

  • The ecosystem is smaller than Unity or Unreal.

Who Should Use GDevelop?

GDevelop is an excellent choice for beginners, indie developers, educators, students, artists, designers, and small teams.

It is especially useful when the goal is to turn a game idea into a working prototype quickly. Teachers can also use it to introduce game development without requiring students to learn a programming language first.

Experienced developers can benefit from the combination of visual events and JavaScript, making GDevelop more flexible than a typical no-code game maker.

Final Verdict

GDevelop has evolved into much more than a simple no-code game maker. Its combination of 2D and 3D capabilities, visual programming, JavaScript support, ready-made behaviors, multiplayer services, AI assistance, and broad export options makes it a surprisingly capable game development platform.

The biggest reason to choose GDevelop is its accessibility. You can start building a playable game without spending weeks learning programming or a complicated engine interface. As your project becomes more advanced, JavaScript and extensions provide an escape hatch when visual events are no longer enough.

For beginners and indie developers, GDevelop is one of the strongest open source game engines available. For experienced developers working on technically demanding 3D productions, more specialized engines may still be a better fit.

GDevelop v5.6.286
Free
Download

Software Informations:
Developer:

Operating System:
Windows / macOS / Linux
Date Added:
2026-11-03T14:07:17.100Z

ABBYY FineReader PDF Standard is a professional PDF editor and optical character recognition software designed to convert scanned documents and image-based PDFs into editable and searchable files. It is suitable for Windows users who need accurate text recognition and structured document output.

This edition focuses on essential PDF editing and OCR features without including enterprise-level collaboration tools found in higher-tier versions.

Get Discount 20% OFF using "SWMR" Coupon code for the ABBY Products below:

ABBYY FineReader PDF Corporate (3-year plan)

ABBYY FineReader PDF Corporate (monthly plan)

ABBYY FineReader PDF Corporate (yearly plan)

ABBYY FineReader PDF for Mac (3-year plan)

ABBYY FineReader PDF for Mac (yearly plan)

ABBYY FineReader PDF Standard (3-year plan)

ABBYY FineReader PDF Standard (monthly plan)

ABBYY FineReader PDF Standard (yearly plan)

ABBYY Screenshot Reader

Key Features of ABBYY FineReader PDF Standard

High Accuracy OCR Technology

One of the biggest strengths of ABBYY FineReader PDF Standard is its advanced OCR engine. It recognizes text from scanned pages, photos, and non-searchable PDFs with impressive accuracy. It also preserves layout elements such as tables, columns, and formatting.

The software supports multiple languages, which makes it suitable for international users.

PDF Editing Capabilities

ABBYY FineReader PDF Standard allows users to edit text and images directly inside PDF files. You can modify paragraphs, adjust fonts, delete content, or insert additional pages without converting the file to another format first.

Convert PDF to Editable Formats

The software enables conversion of PDFs into commonly used formats, including:

  • Microsoft Word

  • Microsoft Excel

  • Microsoft PowerPoint

  • RTF

  • TXT

  • Searchable PDF

Converted documents maintain structure and formatting, reducing the need for manual corrections.

Create Searchable and Archival PDFs

You can convert scanned documents into searchable PDFs for long-term storage. This is useful for businesses managing contracts, invoices, and records.

Organize and Manage PDF Files

ABBYY FineReader PDF Standard provides tools to rearrange pages, merge documents, and manage large PDF files efficiently.

Performance and User Experience

ABBYY FineReader PDF Standard performs smoothly even with large documents. The interface is clean and intuitive, allowing users to access OCR and editing tools without a steep learning curve.

OCR processing is fast, and the text recognition quality is consistently accurate. Layout retention during conversion is one of its strongest advantages compared to many free PDF tools.

Final Verdict

ABBYY FineReader PDF Standard remains one of the most reliable PDF OCR and editing tools in 2026. Its strong text recognition accuracy, structured conversion output, and practical editing tools make it a solid investment for anyone working extensively with documents.

If you are looking for reliable PDF software with advanced OCR and editing features, ABBYY FineReader PDF Standard is one of the most trusted solutions available. In this detailed ABBYY FineReader PDF Standard review, we will explore its features, performance, advantages, and overall value.

Designed for users who regularly work with scanned documents and PDFs, this software combines powerful text recognition with practical editing tools.

For dependable OCR and professional PDF editing, ABBYY FineReader PDF Standard continues to be a top choice.

ABBYY FineReader PDF 16 Release 2026.6.3 Standard (3-year plan) - Discount 10% OFF
$240.3$267
Download

Software Informations:
Developer:

Operating System:
Windows
Date Added:
2026-11-03T12:07:30.923Z

Handy is a free, open-source desktop application for speech-to-text conversion that transforms spoken words into written text, inserting them directly into whatever application you happen to be using. Instead of recording audio and manually transcribing the content via other services, Handy allows you to dictate triggered by a simple keyboard shortcut directly into web browsers, text editors, messaging apps, and any other text input fields.

The application prioritizes simplicity and privacy. Speech recognition is performed locally using artificial intelligence models such as Whisper and Parakeet, this means that, for standard transcription, there is no need to upload your microphone recordings to any cloud-based service.

Developed using Rust and Tauri, Handy is available for Windows, macOS, and Linux. It is highly useful for writers, software developers, students, and anyone who prefers speaking over typing.

Features

Speech-to-Text Anywhere

Handy's primary function is inserting transcribed speech into the currently active text field.

Users can press a configurable keyboard shortcut, speak naturally, and release the key or press the shortcut again to finish recording. Handy processes the audio and types the resulting transcript into the active application.

This makes it convenient for:

  • Writing emails and documents.

  • Composing messages and replies.

  • Entering text into web forms.

  • Dictating notes and technical documentation.

  • Writing content in code editors and other desktop applications.

Unlike conventional transcription applications, Handy does not require users to work inside a dedicated editor.

Multiple Speech Recognition Models

Handy supports different speech recognition models to accommodate different languages, hardware configurations, and transcription requirements.

Whisper is suitable for multilingual transcription and supports more than 99 languages. Users can choose different model sizes depending on their hardware and preferred balance between speed and accuracy.

Parakeet V3 is optimized for CPU-based processing and offers automatic language detection. It is particularly useful for users who want relatively fast transcription without relying on a dedicated GPU.

Cohere Transcribe is another available option, designed for high-quality English transcription. It is larger and more demanding, making it better suited to computers with capable graphics hardware.

Model selection is important because transcription speed, accuracy, language support, and resource consumption vary considerably.

Completely Offline Processing

One of Handy's strongest advantages is local speech recognition. Once the required model has been downloaded, ordinary transcription can run without sending microphone audio or transcript content to an external transcription service.

This is useful for people who regularly dictate private messages, business documents, personal notes, or other sensitive content.

Handy does not require an account or cloud subscription for its basic transcription functionality. Network access is used for activities such as downloading models and application updates, along with optional services that users choose to configure.

Voice Activity Detection

Handy uses Silero Voice Activity Detection to identify speech and filter silence from recordings.

This helps reduce unnecessary audio processing and improves the transcription workflow, particularly when users pause naturally while speaking.

Removing silence can also make the application more convenient for longer dictation sessions.

Push-to-Talk and Toggle Modes

The application provides different recording controls.

In push-to-talk mode, users hold the configured shortcut while speaking and release it when finished. Toggle mode starts recording with one shortcut press and stops it with another.

Keyboard shortcuts can be customized to suit individual workflows and avoid conflicts with existing application shortcuts.

AI Post-Processing

Handy also provides an optional post-processing workflow for cleaning up transcribed text.

This can help improve punctuation, formatting, and readability after speech recognition. Depending on the configuration, post-processing can use an external AI provider, so it should not be confused with the fully local transcription workflow.

Users who prioritize privacy can use ordinary local transcription without enabling an external post-processing service.

Custom Dictionary and Transcription Settings

Handy provides settings for controlling transcription behavior, including model selection, language configuration, keyboard shortcuts, and custom vocabulary.

A custom dictionary is particularly useful when dictating technical terms, product names, abbreviations, or words that speech recognition models might otherwise misinterpret.

These options make the application more adaptable to specialized writing and professional workflows.

Cross-Platform Support

Handy supports Windows, macOS, and Linux.

It integrates with the desktop environment through global keyboard shortcuts and system-level text input. Depending on the operating system, it may require microphone and accessibility permissions.

Linux users may need additional utilities for reliable text insertion, depending on whether they use X11 or Wayland.

Download Handy v0.10.0 - Software Mirrors

Handy v0.10.0 for Windows

Handy_0.10.0_x64_en-US.msi | 31.18 MB

Handy_0.10.0_x64-setup.exe | 16.45 MB

Handy_0.10.0_arm64_en-US.msi | 14.54 MB

Handy_0.10.0_arm64-setup.exe | 10.26 MB

Handy v0.10.0 for macOS

Handy_0.10.0_x64.dmg | 14.39 MB

Handy_0.10.0_aarch64.dmg | 14.09 MB

Handy v0.10.0 for Linux

Handy_0.10.0_arm64.deb | 28.31 MB

Handy_0.10.0_amd64.deb | 32.55 MB

Handy_0.10.0_amd64.AppImage | 118.05 MB

Handy_0.10.0_aarch64.AppImage | 110.49 MB

Handy-0.10.0-1.x86_64.rpm | 86.33 MB

Handy-0.10.0-1.aarch64.rpm | 73.39 MB

Others Download related to Handy v0.10.0

Handy_x64.app.tar.gz | 14.42 MB

Handy_aarch64.app.tar.gz | 14.24 MB

Handy v0.10.0 Source Code

Handy v0.10.0 Source code (zip)

Handy v0.10.0 Source code (tar.gz)

Handy v0.10.0 Release Notes:

This is a substantial release, and it marks Handy being roughly feature complete as we move towards V1. In this release, there are some big features like transcribing audio files, auto language detection for all of the transcription models we support, a brand new onboarding flow, and starting to redesign some of the UI. There are a ton more quality of life improvements and small little details that hopefully you'll enjoy as well. The coming releases will be addressing issues and stability. From here on out, you shouldn't expect too many new features from Handy, but you should expect the app to become more beautiful and stable. Given there's a lot of new features, likely there will be some new issues, and we will address those in the rest of this release, and it will wrap us up and bring us to V1. A massive, massive thank you to all the contributors, for this release. A ton of PRs have came in, and really appreciate people submitting fixes and continuing to help the app improve.

What's Changed

  • Use Handy icon for Windows NSIS installer by @MikaKrul in #2117
  • fix: refresh custom sounds when opening sound selector by @MikaKrul in #1941
  • transcribe 0.3.0 by @cjpais in #2203
  • Docs/ubuntu 26 04 gnome wayland troubleshooting by @henderson-01 in #2206
  • docs: add overlay, wl-clipboard and non-QWERTY paste notes to the Ubuntu notes by @B-Galati in #2212
  • fix(portable): retain access to legacy model cache by @leongrphc in #2116
  • isolate transcribe.cpp backend in own worker by @cjpais in #2208
  • fix: show that the model is loading instead of a stuck "Transcribing..." by @maurild14 in #2227
  • docs: keep translation contribution guidance aligned with supported locales by @johnjheejin in #2235
  • fix(linux): allow renderer opt-out without dropping crash protection by @cjpais in #2249
  • transcribe 0.4.0 by @cjpais in #2265
  • move to handy-recorder by @cjpais in #2188
  • transcribe.cpp vad by @cjpais in #2267
  • feat: support Anthropic Messages API format in the Custom post-processing provider by @diligentpig in #2257
  • fix: avoid duplicate history rows during live updates by @johnjheejin in #2256
  • fix: expose setting names to screen readers by @johnjheejin in #2238
  • add ui to migrate onnx model users to gguf by @cjpais in #2277
  • actually remove onnx by @cjpais in #2279
  • feat: pass custom words via transcribe-cpp vocabulary API by @cjpais in #2284
  • long form chunking with vad by @cjpais in #2282
  • catalog from @handy-computer/models package by @cjpais in #2299
  • fix: stop blocking the main thread during paste by @cjpais in #2302
  • fix: center grade chip labels on Windows by @cjpais in #2326
  • feat: transcribe files into history by @cjpais in #2315
  • add language detection by @cjpais in #2331
  • fix: declare OpenSSL 3 runtime dependency in Debian packages by @mvanhorn in #2336
  • fix: reuse one HTTP client for post-processing requests by @rnmrnmrnmrnmrnmrnm in #2344
  • docs: document stale user manager trap for input-group permission fixes by @haroon0x in #2348
  • fix: preserve dictation when post-processing returns blank text by @johnjheejin in #2290
  • fix: identify the setting each reset button affects by @johnjheejin in #2253
  • fix(shortcuts): avoid ibus Ctrl+Space collision on Linux (#2332) by @Bl0ck154 in #2367
  • fix: let Escape dismiss dropdown menus by @johnjheejin in #2246
  • error notices framework by @cjpais in #2372
  • fix: ignore IME confirmation Enter in text inputs by @johnjheejin in #2248
  • fix: report failed history deletions to users by @johnjheejin in #2255
  • fix: name the audio playback position control by @johnjheejin in #2254
  • fix: preserve playback position when starting a seek by @johnjheejin in #2247
  • fix: dismiss focused setting tooltips with Escape by @johnjheejin in #2251
  • fix: release late audio sources after leaving history by @johnjheejin in #2309
  • fix: restore shortcut fallback when manager startup fails by @utpalsinghdev in #2327
  • fix(macos): prevent silent AirPods recordings when also playing audio from iPhone by @mattdoran in #1991
  • Add transcription hook by @AlexanderYastrebov in #930
  • fix(linux): add 'None' keyboard implementation for Wayland compositors by @aasmall in #1300
  • fix(linux): type multi-line text correctly with dotool by @Huge in #2199
  • feat(tray): add language switching submenu by @eti0 in #1319
  • fix(linux): keep transparent overlay painting on WebKitGTK 2.54 by @xilec in #2360

New Contributors

  • @henderson-01 made their first contribution in #2206
  • @B-Galati made their first contribution in #2212
  • @maurild14 made their first contribution in #2227
  • @johnjheejin made their first contribution in #2235
  • @rnmrnmrnmrnmrnmrnm made their first contribution in #2344
  • @haroon0x made their first contribution in #2348
  • @Bl0ck154 made their first contribution in #2367
  • @utpalsinghdev made their first contribution in #2327
  • @mattdoran made their first contribution in #1991
  • @aasmall made their first contribution in #1300
  • @Huge made their first contribution in #2199
  • @eti0 made their first contribution in #1319
Full Changelog: v0.9.8...v0.10.0

Performance and Compatibility

Handy is built using Rust and Tauri, with a React and TypeScript interface. Audio processing and speech recognition are handled locally through its underlying transcription libraries.

Performance depends primarily on the selected model and available hardware.

Whisper can use GPU acceleration when supported, while Parakeet V3 is designed to work efficiently on CPUs. This makes Handy accessible to users without dedicated graphics cards, although larger models generally require more processing resources.

The application is lightweight compared with full-featured voice assistants because it concentrates on one main task: converting speech into text.

There are some platform-specific limitations worth considering:

  • Windows: Supports x64 and ARM64 releases, with microphone access and global shortcut support.

  • macOS: Supports Intel and Apple silicon Macs. Microphone and Accessibility permissions are required for normal operation.

  • Linux: Supports x64 and ARM64 distributions, but text insertion may require additional utilities such as xdotool, wtype, or dotool, depending on the desktop environment.

Handy is not intended to replace professional audio transcription suites with advanced speaker identification, subtitle editing, or transcription management. Its strength is quick, direct dictation into existing applications.

System Requirements

Handy's requirements depend on the selected speech recognition model.

  • Windows: Compatible with supported 64-bit Windows systems, including x64 and ARM64 builds.

  • macOS: Supports Intel and Apple silicon Macs.

  • Linux: Supports x64 and ARM64 environments.

  • Whisper: Can use compatible Intel, AMD, or NVIDIA hardware, with GPU acceleration when available.

  • Parakeet V3: CPU-only operation is supported, with Intel Skylake or equivalent AMD processors recommended.

  • Storage: Additional space is required for downloaded speech recognition models.

  • Permissions: Microphone access is required, along with appropriate permissions or system utilities for text insertion.

For users with limited hardware, Parakeet V3 or a smaller Whisper model is a reasonable starting point.

Pros and Cons

Pros

  • Completely free and open-source under the MIT License.

  • Runs speech recognition locally without requiring cloud transcription.

  • Inserts transcribed text directly into almost any active text field.

  • Supports Whisper and Parakeet speech recognition models.

  • Supports multilingual transcription through Whisper.

  • Parakeet V3 provides CPU-based processing and automatic language detection.

  • Configurable keyboard shortcuts and push-to-talk controls.

  • Uses Silero for voice activity detection.

  • Optional AI post-processing for improving transcripts.

  • Available for Windows, macOS, and Linux.

  • Does not require an account or subscription for basic functionality.

Cons

  • Transcription speed depends on hardware and model selection.

  • Larger Whisper models require more processing resources.

  • Linux users may need additional packages for reliable text insertion.

  • Some optional post-processing features require external AI services.

  • Does not provide a full transcription editor or advanced audio-management tools.

  • Requires manual model downloads before offline transcription can begin.

How to Install

  1. Visit the official Handy website.

  2. Download the installer for Windows, macOS, or Linux.

  3. Install the application using the downloaded package.

  4. Launch Handy and grant microphone permissions.

  5. Grant Accessibility permissions on macOS when requested.

  6. Select a speech recognition model, such as Whisper or Parakeet V3.

  7. Wait for the model download to finish.

  8. Configure your preferred keyboard shortcut and recording mode.

  9. Open any application with a text field.

  10. Press the shortcut, speak, and finish recording to insert the transcription.

Linux users should check the installation documentation for additional text-input dependencies and display-server compatibility.

Final Verdict

Handy is an excellent lightweight speech-to-text application for users who want convenient voice typing without depending on cloud transcription services. Its ability to insert text directly into existing applications makes it more practical than traditional transcription tools for everyday writing.

Support for Whisper and Parakeet provides flexibility across different languages and hardware configurations. The local processing approach is another major advantage for privacy-conscious users, while its open-source MIT license makes the project accessible to developers who want to inspect, modify, or extend the software.

Its main limitations are the lack of advanced transcription management features and some platform-specific setup requirements, particularly on Linux. Optional AI post-processing also requires additional configuration and may involve external services.

Handy is highly recommended for anyone looking for free, private, and convenient offline voice typing. Its focused design, cross-platform support, and local AI transcription make it a useful everyday productivity tool.

Handy v0.10.0
Free
Download

Software Informations:
Developer:

Operating System:
Windows / Linux / macOS
Date Added:
2026-11-03T12:07:25.558Z
Subscribe to: Posts (Atom)