Craft Anything v0.3.0

Craft Anything is an open-source tool designed to help users understand existing software and create their own versions using AI. Developed by the ArtCraft team, the tool integrates software inspection, reverse engineering utilities, and AI-driven programming into a single workflow.

Rather than simply generating applications from text-based queries, Craft Anything analyzes a target website, desktop application, mobile app, or game. It then uses the gathered information to build a new project, compares the result against the original, and further refines the output.

Features

AI-Powered Software Recreation

Craft Anything follows a four-stage workflow: Study, Build, Check, and Refine. Users provide a target and describe what they want to create. The software examines the target, sends the findings to a coding agent, generates a new project, and checks how closely the result matches the original.

It supports recreating websites, applications, and games. Users can also specify their preferred programming language, framework, or game engine.

Reverse Engineering and Inspection

The underlying engine is written in Rust and includes a large collection of inspection tools for different software formats and platforms.

It can examine native applications, Java and .NET programs, Android packages, Apple applications, Electron applications, websites, firmware, databases, game files, and other formats.

The Advanced workbench exposes tools such as disassembly, decompilation, function analysis, call graphs, hexadecimal inspection, and evidence records. Users can investigate software without necessarily starting a recreation project.

Clean Room and Deep Modes

One of the more interesting features is the separation between Clean Room and Deep modes.

Clean Room is the default mode. It studies software through publicly visible behavior, screenshots, and other permitted observations without reading the original application's source code or files. This approach is intended to reduce the risk of copying protected implementation details.

Deep mode enables more extensive inspection, including reading application files and using additional reverse-engineering tools. Ghidra integration provides further decompilation capabilities when installed.

AI Coding Agent Integration

Craft Anything relies on an external coding agent to build applications. Claude Code is the primary supported integration and runs using the user's existing installation and authentication.

The project also includes adapters for other agents, although some integrations are not yet verified in real-world installations.

For analysis and questions, it supports Anthropic, OpenAI, OpenRouter, and local OpenAI-compatible servers.

CLI and MCP Support

The software is not limited to its desktop interface. It also provides a command-line tool and an MCP server.

This allows compatible AI agents and automation workflows to use the same inspection engine. The project advertises 284 tools across its engine, AI, and setup layers, with each tool accessible through the CLI and MCP interfaces.

Build Verification

Generated applications are not simply accepted after the coding agent finishes. Craft Anything checks the output against the original target.

For websites, it captures and compares layouts at desktop, tablet, and mobile widths. It also identifies differences in layout, typography, spacing, page structure, and other visual details.

Game projects can use their own self-tests to check basic functionality. The system can then ask the coding agent to refine the result when checks identify problems.

Download Craft Anything v0.3.0 - Software Mirrors
Craft Anything v0.3.0 for Windows craft-anything-0.3.0-windows-x86.msi | 50.58 MB craft-anything-0.3.0-windows-x86-portable.zip | 59.47 MB craft-anything-0.3.0-windows-x64.msi | 52.67 MB craft-anything-0.3.0-windows-x64-portable.zip | 62.73 MB craft-anything-0.3.0-windows-arm64.msi | 49.82 MB craft-anything-0.3.0-windows-arm64-portable.zip | 60.22 MB
Craft Anything v0.3.0 for macOS craft-anything-web-0.3.0.zip | 2.53 MB craft-anything-cli-0.3.0-macos-universal.zip | 24.11 MB craft-anything-0.3.0-macos-universal.dmg | 121.94 MB
Craft Anything v0.3.0 for Linux craft-anything-0.3.0-linux-x86_64.tar.gz | 63.56 MB craft-anything-0.3.0-linux-x86_64.rpm | 64.11 MB craft-anything-0.3.0-linux-x86_64.pkg.tar.zst | 49.74 MB craft-anything-0.3.0-linux-x86_64.flatpak | 46.89 MB craft-anything-0.3.0-linux-x86_64.deb | 64.51 MB craft-anything-0.3.0-linux-x86_64.AppImage.zsync | 211.63 kB craft-anything-0.3.0-linux-x86_64.AppImage | 60.39 MB craft-anything-0.3.0-linux-riscv64.tar.gz | 62.45 MB craft-anything-0.3.0-linux-riscv64.rpm | 63.13 MB craft-anything-0.3.0-linux-riscv64.deb | 63.37 MB craft-anything-0.3.0-linux-aarch64.tar.gz | 60.89 MB craft-anything-0.3.0-linux-aarch64.rpm | 62.06 MB craft-anything-0.3.0-linux-aarch64.pkg.tar.zst | 47.81 MB craft-anything-0.3.0-linux-aarch64.flatpak | 43.89 MB craft-anything-0.3.0-linux-aarch64.deb | 62.62 MB craft-anything-0.3.0-linux-aarch64.AppImage.zsync | 203.17 kB craft-anything-0.3.0-linux-aarch64.AppImage | 57.98 MB
Craft Anything v0.3.0 for BSD craft-anything-0.3.0-freebsd-x86_64.tar.gz | 61.17 MB craft-anything-0.3.0-freebsd-aarch64.tar.gz | 58.21 MB
Craft Anything v0.3.0 Source Code Craft Anything v0.3.0 Source code (zip) Craft Anything v0.3.0 Source code (tar.gz)
Craft Anything v0.3.0 Release Notes: What's Changed A .utoc naming billions of empty compression methods no longer runs out of memory by @Agapenordev in #61 Find the Windows shim, not npm's POSIX script, for a bare program name by @muraa-p in #51 Every engine listing finds a Ghidra installed in a versioned folder or set through its folder setting by @aweussom in #55 Leave a pin's old program copies out of the copy by @muraa-p in #56 Look for Claude Code where WinGet installs it by @justinjohn0306 in #65 Analysing code at the top of the address space finishes instead of running forever by @Agapenordev in #78 The macOS app ships craft-anything-cli beside its executable by @bflatastic in #99 Codex runs in project folders that aren't git repositories by @bflatastic in #101 Find Ghidra installed from a Linux package by @bflatastic in #103 Emit object schemas for compare inputs by @TheGalitube in #93 Wait for asset facts to load in the asset spending smoke test by @WadahBadawi in #76 Scope test structs FlakyKind and FlakyCheck to unix by @WadahBadawi in #52 Frida lists devices and processes on Windows instead of hanging or finding none by @vdidon in #59 Add a French UI translation by @VakzOs in #70 Add a Japanese README by @zitann2020 in #80 The Make screen says Craft Anything looks at the original, not an unnamed "we" by @bflatastic in #107 Add a Japanese UI by @zitann2020 in #87 The Agent console keeps its size instead of growing until its tabs are cut off by @Agapenordev in #81 Why we ask is shown on the first-run purpose page, not hidden in a hover tip by @bflatastic in #106 The Settings window keeps its width on every tab by @bflatastic in #108 A rebuilt stand-in waits for its predecessor's port instead of failing by @bflatastic in #109 Look for AI offers local servers already set up in OpenCode by @bflatastic in #111 Fall back to BM_CLICK for off-screen Win32 buttons and check boxes in UI walkthroughs by @WadahBadawi in #77 Sync the Craft App Suite list by @dylanmtaylor in #119 Choose an agent app's program in Settings → AI by @bflatastic in #117 Add Simplified and Traditional Chinese UI translations (R2.9) by @CQMHV in #115 Godot installs on macOS: its program may be bigger than other engines' files by @bflatastic in #118 Find coding agents in OpenCode's install folder and under nvm by @bflatastic in #120 The macOS Dock shows the app's icon at the same size as other apps' by @bflatastic in #121 Translate the agent program setting into Chinese by @bflatastic in #122 Draw CJK interface text in a sans font, not a mincho by @zitann2020 in #125 A program run in a console on Windows doesn't lose what it wrote on a busy computer by @bflatastic in #126 Name the font fallback order, and test the order itself by @zitann2020 in #129 A Flutter library with a header table at the very end of the address range is read without overflowing (R0.5) by @Agapenordev in #131 Look for AI picks a model that can chat, not an embedding model by @bflatastic in #134 Test that a stopped stand-in frees its port where the server is, without racing other tests by @bflatastic in #136 Add an AI by its address in Settings → AI by @bflatastic in #137 Find coding agents npm installed under a custom prefix by @bflatastic in #139 Choose the model an AI uses in Settings → AI by @bflatastic in #145 Find the Godot editor that the store installs on macOS by @olebohmbach in #141 A rate-limited AI service is waited for as long as it asks by @bflatastic in #148 Pitfall: merge the commit CI tested, not a head read later by @bflatastic in #150 Give every Settings tab the same width, with air before the scroll bar by @olebohmbach in #151 A full-page website capture scrolls through the page first by @bflatastic in #168 An out-of-date Claude Code says so, and how to update it by @bflatastic in #169 A website capture refuses the browser's error page by @bflatastic in #171 The README and STATUS give today's tool counts by @bflatastic in #172 When the window can't open, the app says why instead of vanishing by @bflatastic in #173 Browser evidence records Clean room when the run is in Clean room by @bflatastic in #177 A failed build check says where it looked, not app/ for an app edited in place by @bflatastic in #178 A question answered outside a build stays in the project's conversation by @bflatastic in #180 Pitfall: never type python - with a heredoc by @bflatastic in #181 Add Antigravity (agy) agent CLI adapter and setup support by @x-eon-max in #89 Clean room captures leave out class names and query strings by @bflatastic in #183 Setup and the doctor read a client's other names from one table by @bflatastic in #184 Cover the unknown-font rank in the fallback order test by @zitann2020 in #187 A pill corner made with calc(infinity * 1px) is captured as 9999px by @rech4210 in #189 Add a Simplified Chinese README by @maple323 in #188 Detect Unreal Engine installations and Clang on Linux (R1.6) by @one-bit in #191 The Windows app no longer starts OpenGL, which crashed some Intel drivers before the window opened by @bflatastic in #198 Codex is never run as plain text, which opened its interactive mode by @bflatastic in #200 An A32 doubleword load or store into r15 decodes as invalid instead of stopping the ARM analysis by @rech4210 in #204 A Mach-O method list at the end of the address space is read without overflowing by @rech4210 in #205 Reject overflowing bits in 10-byte uleb128 values by @Silvertree2010 in #208 Captured colours are the page's CSS values, not a canvas read that drifted by one by @bflatastic in #211 The Windows pipe gap in project_play is noted in its tool help and kept as an ignored test by @rech4210 in #212 From the CLI, a capture with no open site says each command starts its own browser by @bflatastic in #213 Try the DX12 adapter before Vulkan when the app starts by @muraa-p in #217 "It's done" keeps a build as it is and stops offering to carry on by @bflatastic in #221 A build stops with a plain message when craft-anything-cli is missing by @echelon in #226 Add a Plain HTML and CSS stack so plain web requests use it by @muraa-p in #225 fix(ai): add Hermes Agent CLI adapter and enable ACP support by @vietanh210304 in #16 A same-size rewrite of an unhashed file reads as unknown, not unchanged by @bflatastic in #241 Pseudo-Java keeps the statements before a try by @bflatastic in #242 ELF files with no section headers show their dynamic symbols and functions by @rech4210 in #233 First-run steps scroll when the window is too short, so Continue stays reachable by @bflatastic in #243 The roadmap and status say how approvals differ between agents by @echelon in #244 A capture's timeout stops every process in the group, not just the leader by @bflatastic in #246 adb_app_show says a missing package isn't installed by @bflatastic in #258 The README says what to do when macOS blocks the command-line zip by @echelon in #265 A failed adb download or capture doesn't leave a folder that blocks the retry by @bflatastic in #266 The README's macOS CLI commands work when pasted as they are by @echelon in #267 Setup creates Pi's mcp.json when it doesn't exist yet by @bflatastic in #268 Comparing app maps notices an edited inline script by @bflatastic in #269 android_package_show's summary says signatures are verified by @bflatastic in #290 Saving a key never replaces a key file it couldn't read by @bflatastic in #293 Changing AI settings never saves over ones that couldn't be read by @bflatastic in #295 Resolve Objective-C chained offsets past PAGEZERO (R0.5) by @oliver-mee in #298 Add a Korean README by @rech4210 in #310 Saves from two threads at once never collide or mix a project file by @bflatastic in #311 A large approved action keeps its allow decision over MCP by @bflatastic in #314 An ACP chat starts the agent's session again after a failed start, and Stop ends a waiting message by @bflatastic in #320 The CLI's capabilities list keeps Clean room's refusal of target tools by @bflatastic in #323 The approval check allows npx vite build and still refuses dev servers by @bflatastic in #328 Recognise selected macOS bundles and nested XPC services (R0.5) by @oliver-mee in #297 Enabling a toolset in Clean room counts only the tools it lists by @bflatastic in #324 Catch the Japanese README up with the English one by @zitann2020 in #312 Say the English README wins in the Chinese one too by @zitann2020 in #313 Cached prompt tokens from OpenAI-style APIs are counted once by @bflatastic in #337 Comparing two runs tells a deleted file from a created one by @bflatastic in #340 Comparing folders or archives moves each new file at most once by @bflatastic in #343 The change summary counts a comparison where files only moved as changed by @bflatastic in #345 Import maps, speculation rules and data blocks aren't read as JavaScript by @bflatastic in #353 Bump the version to 0.2.1 by @echelon in #382 Stop advising report --format in the CLI help by @Silvertree2010 in #419 SQLite rowid alias columns show the rowid instead of NULL by @rech4210 in #409 Truncated sBPF .text sections decode readable bytes and report partial coverage by @phatloi0010 in #381 SquashFS extended regular file inodes read their 64-bit layout by @phatloi0010 in #384 Bump the version to 0.3.0 by @echelon in #425 New Contributors @Agapenordev made their first contribution in #61 @muraa-p made their first contribution in #51 @aweussom made their first contribution in #55 @justinjohn0306 made their first contribution in #65 @TheGalitube made their first contribution in #93 @WadahBadawi made their first contribution in #76 @vdidon made their first contribution in #59 @VakzOs made their first contribution in #70 @zitann2020 made their first contribution in #80 @CQMHV made their first contribution in #115 @olebohmbach made their first contribution in #141 @x-eon-max made their first contribution in #89 @maple323 made their first contribution in #188 @one-bit made their first contribution in #191 @Silvertree2010 made their first contribution in #208 @vietanh210304 made their first contribution in #16 @oliver-mee made their first contribution in #298 Full Changelog: v0.2.0...v0.3.0

Performance and Compatibility

Craft Anything is built with Rust and uses its own inspection engine rather than depending entirely on external reverse-engineering utilities. Ghidra can be integrated for additional analysis in Deep mode.

The desktop application, CLI, and MCP server provide different ways to access its capabilities. Its supported release packages cover Windows, macOS, Linux, and FreeBSD, with a browser-based WebAssembly viewer also available for certain inspection tasks.

However, the project is still in early development. Some agent integrations, application behavior checks, and inspection tools are incomplete or have only been tested with limited inputs. The availability of a feature does not necessarily mean every supported software format has been thoroughly validated.

System Requirements

Craft Anything requires Rust 1.97 or newer when building from source.

On Debian and Ubuntu, compiling the desktop application also requires development libraries for X11, Wayland, GTK, and OpenGL.

Building applications with AI requires a compatible coding agent installed and authenticated on the computer. Additional AI providers may require their own API configuration.

Prebuilt installers are available for supported platforms, so users do not necessarily need to compile the application themselves.

Pros and Cons

Pros

• Combines software inspection and AI-powered project creation.

• Supports websites, desktop applications, mobile apps, and games.

• Provides Clean Room and Deep inspection modes.

• Includes extensive Rust-based reverse-engineering tools.

• Supports CLI and MCP integrations.

• Compares generated websites against original layouts.

• Open-source under MIT or Apache-2.0 licensing.

Cons

• Still in early alpha development.

• Requires an external coding agent for building projects.

• Some agent integrations and verification features are incomplete.

• Deep analysis can require additional tools and substantial system resources.

• Results depend on the quality of the target inspection and the coding agent.

How to Install

The easiest approach is to download a prebuilt package from the project's GitHub Releases page.

Available formats include MSI installers for Windows, DMG for macOS, and AppImage, Flatpak, DEB, and RPM packages for Linux.

Users who prefer compiling from source can clone the repository and run:

git clone https://github.com/storytold/craft-anything
cd craft-anything

cargo run --release -p craft-anything

The CLI can be installed separately:

cargo install --path apps/craft-anything-cli

After installation, run the diagnostic command to check the environment:

craft-anything-cli doctor

Final Verdict

Craft Anything is an ambitious open-source project that combines reverse engineering with AI-assisted software development. Its biggest advantage is the complete workflow: inspect an existing application, understand how it behaves, generate a new implementation, and check the result against the original.

The Clean Room mode is particularly interesting because it provides a more controlled approach to recreating software based on observable behavior rather than directly copying its implementation. Meanwhile, Deep mode gives experienced developers access to more extensive analysis capabilities.

Its main limitation is maturity. The project is still in early alpha, and some integrations and verification features have not been fully validated. Users should treat generated results as development starting points rather than guaranteed replacements for existing software.

Overall, Craft Anything is a promising tool for developers interested in reverse engineering, software prototyping, and AI-powered application recreation. It is worth exploring, especially for technically experienced users who want more control over how AI studies and rebuilds software.

Sources

• Craft Anything GitHub Repository

• Craft Anything Releases

• Craft Anything Development Status

Free

Download

Software Informations:
Developer:
ArtCraft
Operating System:
Windows / macOS / Linux / BSD
Date Added:
2026-11-03T04:02:04.247Z

Hex v3.3.0 - Private, Local AI Voice Dictation

Developed by Anomaly, Hex is an open-source voice dictation application for desktop computers. It converts speech into text and inputs it directly into the application the user is currently using. Instead of opening a separate transcription editor, users simply hold down a keyboard shortcut while speaking; once the key is released, the recognized text is pasted into the active application.

Built using the Rust programming language, Hex utilizes locally downloaded speech recognition models. This means it does not rely on cloud services to function. It is compatible with macOS, Windows, and Linux; however, according to the project documentation, the Windows and Linux versions are still in beta. Hex is highly useful for software developers, writers, and anyone who frequently needs to input text into web browsers, terminals, code editors, or messaging applications.

Hex v3.3.0 - Private, Local AI Voice Dictation Release Notes:

Hex 3.3.0

new

• the first run has a quiet skip on the demo and practice, and if the speech model download fails or stalls you can carry on and pick a model later

• the speech model picker compares three models side by side, then shows every model

• russian: gigaam v3, and you can add your own gigaam model files

• the interface speaks traditional chinese (繁體中文)

• shortcut mode is one choice in general, and hex only offers the modes your desktop can deliver

• each mode can have its own vocabulary

• hex explains the desktop's paste permission before the desktop asks for it

• when you set up a newer hex, an older copy that's still running quits by itself

• mac: organizations can lock features on or off with a configuration profile

• linux: apps can start and stop dictations through hex's local api, like on mac

• new in labs: skip long pauses before transcribing (faster on slow computers), stop a meeting recording 30 s after the call ends, and fix a wrong result right from output's try box

fixes

• mac: the shortcut works again after cmd+tab or cmd+q left a key looking held, after hex sat idle for a while, and when it's fn

• windows: dictation starts on mics that got stuck on "enabling…", and stream deck and other injected shortcut presses no longer get cut to nothing

• dictation no longer fails with "the current clipboard format cannot be preserved safely"

• spoken lists and numbers stay as you said them when hex cleans up your text

• clearing or deleting history sticks, even after a failed try

• if your opencode 2 is too old for hex, hex tells you to update it instead of skipping the ai rewrite without a word

• linux: hex picks up settings you edit outside the window, the status and indicator name the mic in use and warn when it goes away, and hex no longer crashes at startup on gnome + nvidia under xwayland

• kde: hex shows up in the shortcuts settings, waits for you to approve the shortcut, and asks again if you didn't. the default is ctrl+alt+space

• gnome: the dictation shortcut registers reliably

• niri and other desktops whose shortcuts portal can't bind: hex still starts listening, and hex dictation works from your own keybinding

• annotate: the pen works on hyprland and on fractional scaling, saved ink lines up, recording works on intel + nvidia laptops, and captures finish cleanly

• custom models that failed to warm up on the gpu load again

• the speech engine row no longer says the gpu stopped responding when it's fine

• meetings: the ready and recording capsules have a close button

Features

System-Wide Voice Dictation

Hex works across desktop applications rather than restricting transcription to its own interface.

The basic workflow is straightforward:

1. Hold the configured keyboard shortcut.

2. Speak into your microphone.

3. Release the shortcut.

4. Hex transcribes the recording and pastes the text into the active application.

This approach makes it convenient for writing emails, entering commands, composing messages, documenting code, and filling out web forms.

Users can configure the shortcut to fit their workflow and avoid conflicts with other applications.

Local AI Speech Recognition

Hex downloads speech recognition models during initial setup and runs transcription locally on the computer.

This reduces dependence on external transcription services and allows ordinary dictation to work without continuously uploading microphone recordings.

The project supports multiple speech recognition models, including Whisper and Parakeet-based models, with additional model options available as the project evolves.

Different models offer different trade-offs in language support, accuracy, memory consumption, and processing speed.

Multiple Languages

Hex supports multilingual dictation through its speech recognition models. Whisper is particularly useful for users who need transcription in languages beyond English.

Model selection matters because not every model supports the same languages. Users should check the available model descriptions before choosing one for regular dictation.

Modes and Application-Specific Behavior

Hex includes Modes that allow users to customize how dictated text is handled.

These can support different writing contexts, such as:

• General writing and everyday messages.

• Technical documentation.

• Programming-related text.

• Application-specific vocabulary and formatting.

• Text correction and optional post-processing.

This is useful because dictated text for an email does not necessarily need the same formatting as text intended for a terminal or code editor.

Some processing modes can use external providers, so optional AI rewriting should be distinguished from ordinary local transcription.

Custom Vocabulary

The vocabulary feature helps improve recognition of names, technical terminology, abbreviations, and other words that speech recognition models might otherwise misinterpret.

This is particularly useful for developers who regularly dictate programming terms, command names, product names, and specialized terminology.

Voice Activity and Recording Controls

Hex includes audio controls designed to make dictation more convenient.

Its implementation supports releasing the microphone while idle and controlling audio from other applications during dictation on supported platforms.

These features help reduce unnecessary microphone activity and make the recording experience more comfortable.

History and Transcript Management

Hex maintains a history of previous dictations, allowing users to revisit recognized text rather than relying exclusively on immediate insertion.

This is useful when users need to recover a previous transcription or check what was recognized during an earlier session.

Cross-Platform Desktop Support

Hex provides desktop builds for macOS, Windows, and Linux.

Its Linux version supports local hotkey dictation, Modes, History, and signed updates. Windows builds are available for both x64 and ARM64 systems.

Platform-specific requirements apply, particularly to global shortcuts, graphics acceleration, accessibility permissions, and text insertion.

Downloads

Stable is the current release. Dev is the newest build of dev and updates itself from the dev channel.

Platform	Stable	Dev
macOS (Apple silicon)	DMG	DMG
Windows x64	Installer	Installer
Windows ARM64	Installer	Installer
Linux x86_64	Executable	Executable
Linux x86_64 (Debian, Ubuntu)	.deb	.deb
Linux x86_64 (Fedora, openSUSE)	.rpm	.rpm
Linux ARM64	Executable	Executable
Linux ARM64 (Debian, Ubuntu)	.deb	.deb
Linux ARM64 (Fedora, openSUSE)	.rpm	.rpm

Or install with:

• Homebrew: brew install --cask anomalyco/tap/hex

• AUR: yay -S hex-bin

• install script (Linux): curl -fsSL https://anoma.ly/hex/install-linux.sh | sh

• Nix: nix profile install github:anomalyco/hex

More docs are in docs/.

Performance and Compatibility

Hex is designed to keep the dictation workflow responsive without requiring a dedicated cloud transcription service.

Its native Rust implementation handles audio processing and model execution. On supported x64 systems, Vulkan GPU acceleration can improve transcription speed. CPU-based processing is also available on supported configurations, although performance depends on the selected model and hardware.

Supported platforms include:

• macOS: Apple silicon Macs running macOS 15 or newer.

• Windows: Windows x64 and ARM64. Windows support is currently beta.

• Linux: x86_64 and ARM64 distributions. Linux support is currently beta.

Linux compatibility requires additional attention. Global shortcuts, clipboard access, and text insertion depend on desktop environment capabilities. Some Wayland configurations require compatible compositor protocols or additional utilities.

Hex is best suited to quick dictation rather than professional transcription management. It does not aim to replace dedicated transcription suites with extensive editing, subtitle synchronization, or speaker-management tools.

System Requirements

Requirements depend on the operating system and selected speech recognition model.

• macOS: Apple silicon and macOS 15 or newer.

• Windows: Windows x64 or ARM64. The x64 build uses Vulkan for GPU processing and requires a compatible NVIDIA, AMD, or Intel graphics driver. ARM64 uses CPU processing.

• Linux: Compatible x86_64 or ARM64 environment, with Vulkan acceleration available on supported hardware.

• Storage: Additional disk space for downloaded speech recognition models.

• Microphone: A working microphone and appropriate operating system permissions.

• Internet: Required for downloading the application, speech models, and updates. Ordinary local transcription can subsequently work offline.

The application requirements should not be confused with the requirements of individual models. Larger models can consume significantly more memory and processing resources.

Pros and Cons

Pros

• Free and open-source under the MIT License.

• Local speech recognition without mandatory cloud transcription.

• Inserts dictated text directly into the active application.

• Supports macOS, Windows, and Linux.

• Supports multilingual transcription through compatible models.

• Provides configurable keyboard shortcuts.

• Includes Modes for different writing contexts.

• Supports custom vocabulary.

• Maintains dictation history.

• Native Rust implementation.

• Vulkan acceleration on supported hardware.

• Does not require a subscription for basic local dictation.

Cons

• Windows and Linux versions are still beta.

• macOS support currently targets Apple silicon and macOS 15 or newer.

• Larger speech recognition models require additional resources.

• Linux setup can require additional configuration for Wayland and text insertion.

• Optional AI processing may involve external providers.

• Not designed for advanced transcription editing or meeting management.

• Speech recognition accuracy varies with microphone quality, background noise, and model selection.

How to Install

1. Visit the official Hex website.

2. Select the download package for your operating system.

3. Install the application using the downloaded installer.

4. Launch Hex and complete the initial setup.

5. Download a speech recognition model.

6. Grant microphone and any required accessibility permissions.

7. Configure your preferred keyboard shortcut and transcription language.

8. Open a text editor, browser, or other application.

9. Hold the configured shortcut, speak, and release it to insert the transcription.

Linux users should consult the installation documentation for their desktop environment, particularly when using Wayland.

Final Verdict

Hex is a compelling voice dictation application for users who want to replace repetitive typing with local AI speech recognition. Its biggest advantage is its straightforward workflow: speak using a keyboard shortcut and have the resulting text appear directly in the application being used.

Its local-first architecture is particularly attractive for privacy-conscious users. The combination of multiple speech models, custom vocabulary, application-specific Modes, and cross-platform support makes it more flexible than a basic voice typing utility.

The main drawbacks are its beta status on Windows and Linux, platform-specific requirements, and dependence on local hardware for transcription performance. Optional AI processing also needs separate privacy consideration because not every feature necessarily runs locally.

Hex is an excellent option for developers, writers, and everyday users who want free, private, and convenient AI-powered voice typing. Its focused design and local processing make it a strong alternative to cloud-dependent dictation services.

Sources

• Official product website: Hex

• Source code and documentation: GitHub Repository

• Windows installation and compatibility: Windows Documentation

• Linux implementation and compatibility: Linux Documentation

• Project license: MIT License

Free

Download

Software Informations:
Developer:
Anomaly
Operating System:
Windows / macOS / Linux
Date Added:
2026-11-03T04:01:42.679Z

OpenMausBot v0.1.139

OpenMausBot is an open-source desktop application that allows users to manage various AI agents through an interface reminiscent of a messaging app. The project, formerly known as OpenGrokBot, serves as an alternative to centralized AI personal assistants. It enables users to integrate their own AI tools and organize them into a collection of specialized bots.

Each bot can have its own distinct personality, a specific underlying model, conversation history, computing environment, and connected applications. Instead of relying on a single, universal assistant, users can create agents dedicated to specific tasks such as software development, research, or system administration, among other areas.

OpenMausBot supports Claude, Codex, and Grok models via their respective local command-line interfaces. Furthermore, the application allows for the use of custom tools, connections to OpenAI-compatible endpoints, and various other integration methods. It is currently available for Windows, macOS, and Ubuntu (via the AppImage format).

Features

Multiple AI Agents

OpenMausBot allows users to create multiple AI bots, each with its own instructions, model, and conversation history. Users can switch between agents through a sidebar that works similarly to a messaging application.

This approach makes it easier to organize different responsibilities. For example, one bot can handle programming tasks, another can research technical documentation, and another can assist with project management.

Users can also change an agent's model during an existing conversation, depending on the available provider integrations.

Local AI Agent Execution

One of OpenMausBot's main advantages is its local-first architecture. It uses a local harness server to manage agent processes and communicate with the desktop interface.

The application integrates with installed Claude, Codex, and Grok command-line tools, allowing users to use their existing provider accounts and subscriptions. Custom CLI binaries and compatible engines can also be configured.

Conversation transcripts, credentials, and runtime events are stored locally under the OpenMausBot configuration directory rather than being managed entirely by a centralized cloud service.

However, local execution does not mean every operation stays offline. Cloud AI providers, connected applications, and remote computer environments still require external services.

Dedicated Computer Environments

Each agent can work with its own computer environment. OpenMausBot supports cloud Linux desktops, isolated local virtual machines, and native computer control on supported platforms.

The computer panel provides a live screen preview and an option to open the remote desktop in a browser. This allows users to observe an agent's work and interact with its environment.

Native computer control is platform-dependent. macOS supports the feature, while Ubuntu Xorg requires explicit opt-in and remains in beta. Ubuntu Wayland currently disables host control because of an unresolved safety limitation.

Permission and Approval System

OpenMausBot includes a permission broker that presents certain actions as approval cards. Users can allow or deny shell commands, file modifications, and other requests directly from the conversation.

This is an important feature for agents that can interact with files and computer environments. It gives users more visibility into actions instead of allowing every operation to happen silently.

Nevertheless, users should not assume that every interaction requires approval or that the application provides complete isolation. Review the security documentation and understand the permissions granted to each agent before connecting important files or accounts.

Connected Applications

Through Composio, OpenMausBot can connect agents to external applications such as Gmail, Slack, GitHub, Notion, and Linear, along with hundreds of other supported integrations.

Connections use OAuth-based authentication, allowing agents to interact with approved services without requiring users to manually provide passwords to each bot.

This makes OpenMausBot useful for workflows that involve more than coding. Agents can assist with email, project management, issue tracking, and other connected services.

Composio requires its own account and configuration, and some integrations may have usage limitations or additional costs.

Channels and Team Organization

Channels help organize conversations around different projects or contexts. Each channel can have its own transcript, instructions, working directory, agent roster, and responder rules.

Users can create separate environments for work, personal projects, and individual development tasks without duplicating every agent.

The application also supports importing complete AI teams from Markdown files. These packages can include agent profiles, instructions, channels, playbooks, and suggested routines, making it easier to reuse predefined workflows.

Agent Collaboration

OpenMausBot supports communication between agents and allows them to coordinate work through shared channels.

A primary agent can delegate tasks to other bots, collect their responses, and coordinate a larger workflow. This is useful for projects that benefit from different specialized roles, such as implementation, testing, research, and review.

The effectiveness of collaboration still depends on the selected models and task instructions. Running multiple agents can also increase resource consumption and AI usage.

Scheduling and Webhook Automation

The application includes triggers for recurring tasks and event-driven workflows. Users can configure schedules, webhooks, and connected-app events to start an agent automatically.

Examples include generating regular inbox summaries, responding to incoming events, and running recurring research tasks.

Triggers can be managed through the application's interface, where users can inspect their schedules, pause automations, and review execution history.

However, OpenMausBot must remain available to process local webhook deliveries. Its webhook receiver is not an always-on hosted service, so external access requires an appropriate relay or network configuration.

Voice Interaction

OpenMausBot supports spoken responses through several voice providers, including ElevenLabs, Fish Audio, and Grok. It also supports built-in Mac voices and a local Chatterbox server.

Users can assign different voices to individual agents, making it easier to distinguish between bots during spoken interactions.

Voice calls and native dictation have platform limitations. The current call functionality is macOS-only, while some hosted voice features require separate API credentials.

MCP and Command-Line Integration

OpenMausBot includes a Model Context Protocol server that allows compatible external clients, such as Claude Desktop and Cursor, to interact with its agent workspace.

The MCP integration exposes operations for inspecting bots and channels, searching conversation history, creating tasks, sending messages, changing models, and interrupting agent execution.

It also provides a command-line interface for automating workflows and running tasks from scripts or CI environments. This makes the project more flexible than a conventional desktop chat application.

Performance and Compatibility

OpenMausBot uses Electron for its desktop application, with React, Vite, and Tailwind CSS powering the interface. A separate local harness server manages agent processes and streams events to the application.

Its performance depends heavily on the connected AI models, local hardware, and selected computer environment. The desktop interface itself is relatively lightweight compared with the resources required by active AI agents and virtual machines.

The application supports:

• macOS: Apple silicon and Intel processors.

• Windows: 64-bit Windows systems.

• Linux: Ubuntu 24.04 x64, distributed through DEB and AppImage packages.

Windows and macOS have released desktop builds, while Ubuntu remains in beta. Some computer-control capabilities are restricted depending on the Linux display server.

OpenMausBot is particularly appealing to users who already work with command-line AI agents and want a unified interface for managing them. However, users who expect a completely self-contained application should understand that provider authentication, command-line tools, connected services, and optional cloud computers require separate setup.

System Requirements

The requirements depend on the operating system and the AI tools being used.

• macOS: Apple silicon or Intel Mac.

• Windows: 64-bit Windows desktop environment.

• Linux: Ubuntu 24.04 x64, with GNOME as the documented supported desktop environment.

• Development environment: Node.js 24 or newer and pnpm when running from source.

• AI providers: At least one supported agent CLI, such as Claude, Codex, or Grok, installed and authenticated.

Additional credentials are optional for basic local agent functionality but required for certain integrations, including Composio, Boat cloud computers, and hosted voice providers.

Pros and Cons

Pros

• Open-source desktop application under Apache 2.0, excluding the separately licensed enterprise directory.

• Manages multiple AI agents from one interface.

• Supports Claude, Codex, and Grok through local CLI tools.

• Allows custom engines and compatible model endpoints.

• Provides dedicated computer environments for agents.

• Includes permission cards for reviewing sensitive operations.

• Integrates with hundreds of external applications through Composio.

• Supports agent collaboration, channels, and task delegation.

• Includes scheduling, webhooks, and event-driven automation.

• Provides MCP and CLI interfaces for external automation.

• Stores conversations and configuration locally.

• Supports Windows, macOS, and Ubuntu.

Cons

• Ubuntu desktop support remains in beta.

• Some computer-control features are unavailable on Ubuntu Wayland.

• Requires separate AI provider authentication and command-line tools.

• Cloud computers and connected applications may introduce additional costs.

• Some voice and calling features are platform-dependent.

• API keys entered through the setup process are stored as plaintext in the local configuration file.

• The project is still under active development, so some capabilities may change.

How to Install

1. Visit the official OpenMausBot GitHub repository.

2. Open the Releases section and download the package for your operating system.

3. Install the application using the downloaded installer.

4. Launch OpenMausBot and complete the initial setup.

5. Authenticate at least one supported AI provider, such as Claude, Codex, or Grok.

6. Create your first bot and configure its personality, instructions, and model.

7. Add additional agents for other tasks or projects.

8. Configure optional integrations such as Composio or cloud computer access.

9. Review the permission and security settings before allowing agents to interact with important files or connected accounts.

For developers who want to build from source, the repository provides a pnpm-based workflow:

git clone https://github.com/milind-soni/OpenMausBot
cd OpenMausBot

pnpm install

pnpm dev:server
pnpm dev
pnpm dev:desktop

Node.js 24 or newer, pnpm, and at least one authenticated agent CLI are required.

Final Verdict

OpenMausBot is an interesting open-source project for users who want to move beyond a single AI assistant and build a team of specialized agents. Its combination of local agent execution, dedicated computer environments, connected applications, scheduling, and agent collaboration makes it a flexible workspace for more advanced AI workflows.

The project also stands out through its MCP and CLI integrations, which allow developers to incorporate the workspace into external tools and automation pipelines.

Its main limitations are the ongoing development status, platform-specific restrictions, and the need to configure external AI providers and services. Security also deserves attention, particularly when storing API credentials and allowing agents to control computers or access connected accounts.

OpenMausBot is a promising choice for developers and AI enthusiasts who want an extensible multi-agent workspace with local-first architecture. Its feature set goes beyond ordinary chatbot interfaces, although users should evaluate its security model and platform limitations before adopting it for important workflows.

Sources

• OpenMausBot GitHub Repository

• Official README and Feature Documentation

• OpenMausBot Licensing Information

• OpenMausBot Releases

• OpenMausBot Official Website

Download OpenMausBot v0.1.139 - Software Mirrors
OpenMausBot v0.1.139 for Windows OpenMausBot-setup.exe | 248.14 MB OpenMausBot-0.1.139-setup.exe | 248.14 MB
OpenMausBot v0.1.139 for macOS OpenMausBot.dmg | 296.88 MB OpenMausBot-intel.dmg | 313.32 MB OpenMausBot-0.1.139-x64.zip | 314.51 MB OpenMausBot-0.1.139-x64.dmg | 313.32 MB OpenMausBot-0.1.139-arm64.zip | 297.7 MB OpenMausBot-0.1.139-arm64.dmg | 296.88 MB
OpenMausBot v0.1.139 for Linux OpenMausBot.AppImage | 299.74 MB OpenMausBot-amd64.deb | 254.03 MB OpenMausBot-0.1.139-x86_64.AppImage | 299.74 MB OpenMausBot-0.1.139-amd64.deb | 254.03 MB
Others Download related to OpenMausBot v0.1.139 openmausbot-0.1.139.tgz | 16.21 MB
OpenMausBot v0.1.139 Source Code OpenMausBot v0.1.139 Source code (zip) OpenMausBot v0.1.139 Source code (tar.gz)
OpenMausBot v0.1.139 Release Notes: Lighter and quieter. A workspace with unread bots no longer burns CPU while nothing is happening, the app loads about a third less code at launch, and the server sleeps when nothing is due. Quiet when idle: unread mascots hold still, and the server stops waking up several times a second. Loads less: only your language loads, the rail is its own small page, and the panels you rarely open load when you open them. A screen's first open can take a moment longer, up to about a second on a Cloud page right after it starts. Fewer redraws and writes: long chats reopen faster, messages stop redrawing when something unrelated changes, and the server writes to disk far less while a bot works. Small servers: idle engines close sooner on machines with 1.5 GB of memory or less, and Node's memory limit follows the machine's. Local models: tool calls work even when they arrive as text, a model that loops gets stopped, Ollama gets a big enough context window, and bots on local models get the short tool list. Very long chats still use a lot of memory when scrolled to the top. That's next. What's Changed Other changes fix(memory-upkeep): write memory-upkeep.json once per tick, not once per bot by @ppjsw in #4128 Chat engine: servers past the tool limit stay reachable through the directory, and the bot shows its tool budget by @milind-soni in #4136 OpenCode: refresh re-reads the provider's models; Open at login shows what Windows has by @milind-soni in #4138 fix(chat): wrap long unbroken text inside the user's own bubble by @wlals0874 in #4139 fix(remote): offer full access first when pairing a self-hosted server by @jagtapazad in #4132 docs: explain how to reproduce selected CI test jobs by @garvrao80 in #4113 perf(desktop): load only the shown language, and give the window's rail its own page by @milind-soni in #4144 fix(acp): fail a turn whose whole reply is a provider error body by @plotarmordev in #4114 fix(browser): let the desktop browser through fake-IP proxies (198.18.0.0/15) by @AdityaUmale in #4112 fix(automations): Enter that picks an IME candidate no longer sends the new automation request by @hadan8977 in #4126 fix(moves): a lent computer keeps taking its own screen lease while frozen by @xunguangzlj-cloud in #4125 fix(computer): don't offer This computer on a remote server's page by @wlals0874 in #4137 test(memory): count upkeep writes by file name, so Windows paths match by @milind-soni in #4153 fix(notify): one "done" when a job is really finished by @jagtapazad in #4140 Fetch long tool outputs on open so heavy conversations switch fast by @AdityaUmale in #4141 feat(sidebar): mark a whole department as read by @jagtapazad in #4131 feat(i18n): complete the Simplified Chinese catalog (7288/7288) by @AndersHsueh in #4142 fix(models): mark models the organization turned off, and say when another model answered by @milind-soni in #4172 Bots add, change, sign in to and remove MCP servers from chat by @aivsomkar in #4143 A Chief can answer, for the person, a question a teammate asks them by @darkbringer1 in #4120 Team Clouds enforce their team's rules, fetched from the Admin by @milind-soni in #4176 perf(desktop): quiet when idle — unread mascots hold still, small mascots at 20 fps, stepped compositor sheen by @milind-soni in #4149 Team access: a familiar sharing dialog, and Bot access as a list that opens it by @milind-soni in #4167 A computer frozen for a Cloud that is gone comes home by itself by @milind-soni in #4158 Light look: near-white chrome, a neutral sidebar field, hairline pane edges and one radius scale by @milind-soni in #4150 The server sleeps when nothing is due by @milind-soni in #4156 No config file read per frame, screens only for viewers by @milind-soni in #4155 Write less on the server while a bot works by @milind-soni in #4151 Stop re-rendering every message row when a store value changes by @milind-soni in #4166 Draw each finished message from one Markdown rendering by @milind-soni in #4171 perf(desktop): load less at launch, panels when they open by @milind-soni in #4173 perf: the server fits a small box by @milind-soni in #4177 fix(models): find Claude's account config in the Windows home, and fix the test main fails on by @milind-soni in #4178 Stop asking to use this computer when the bot already works on it by @AdityaUmale in #4157 fix(pages): say publishing isn't available on self-hosted servers instead of asking to sign in by @AdityaUmale in #4107 fix(devs): a department named devs is tagged, not the bug report by @AdityaUmale in #4091 fix: later starts read engines again, and a reloaded engine reads its models by @milind-soni in #4183 feat(agents): let bots read their AI accounts' usage limits by @milind-soni in #4175 fix(usage): show a Claude plan's monthly spend, with its amounts by @milind-soni in #4174 Show spend by department, and count routine and delegated work toward the person by @jagtapazad in #4165 Record bot access changes, answered access requests and owner changes in Activity by @jagtapazad in #4162 Bots on local models get the short tool list without a setting by @milind-soni in #4182 A local model's tool calls work even when they arrive as text, and it can't loop forever by @milind-soni in #4180 GPT models that refuse tools over Chat Completions now use the Responses API by @milind-soni in #4184 fix(electron): soft-fail cloud-send:teams outside My Cloud by @2023Anita in #4170 chore(release): bump version to 0.1.139 by @github-actions[bot] in #4185 Ollama models get a context window big enough for a bot by @milind-soni in #4181 The window is named MausBot before the brand file answers by @milind-soni in #4186 fix(page): the page opens with the product's name as its title by @milind-soni in #4187 Tests: fakes vary their looping calls, answer only POSTs, and match Windows paths by @milind-soni in #4197 Tests: the remaining fakes answer the Ollama probe and expect the local-model rules by @milind-soni in #4206 New Contributors @ppjsw made their first contribution in #4128 @wlals0874 made their first contribution in #4139 @garvrao80 made their first contribution in #4113 @xunguangzlj-cloud made their first contribution in #4125 @AndersHsueh made their first contribution in #4142 @2023Anita made their first contribution in #4170 Full Changelog: v0.1.138...v0.1.139 npm: npx [email protected] · https://www.npmjs.com/package/openmausbot/v/0.1.139

undefined

Free

Download

Software Informations:
Developer:
MausBot community
Operating System:
Windows / macOS / Linux
Date Added:
2026-11-03T04:01:25.863Z

WordCraft v0.15.0

WordCraft is a free, open-source word processor designed as an alternative to Microsoft Word. Developed by the ArtCraft team, it is written in Rust and aims to reproduce familiar Word workflows while providing native applications for desktop operating systems and a browser-based version.

The application goes beyond basic text editing. It includes document formatting, styles, tables, page layout, references, comments, tracked changes, mail merge, and PDF export. It can open and save DOCX files, making it a potential option for users who regularly exchange documents with Microsoft Office users.

WordCraft also includes command-line tools and Model Context Protocol (MCP) integration, allowing compatible AI agents to create, inspect, edit, and export documents programmatically.

Although its feature set is extensive, WordCraft is still approaching maturity for everyday writing. Users who rely on complex Word documents or specialized publishing workflows should test compatibility before making it their primary word processor.

Features

Document Editing and Formatting

WordCraft provides a familiar ribbon-style interface with formatting controls for fonts, sizes, bold and italic text, underlining, colors, highlighting, superscript, and subscript.

It supports paragraph alignment, indentation, line spacing, tabs, borders, shading, and other layout options. The Format Painter lets users copy formatting between document elements, while built-in styles help maintain consistent formatting across longer documents.

The editor also includes AutoCorrect, smart quotes, automatic list formatting, find and replace with regular expressions, and standard undo and redo functionality.

These features make WordCraft suitable for writing reports, letters, essays, and other everyday documents.

DOCX Compatibility and File Formats

WordCraft supports reading and writing Microsoft Word's DOCX format. It also supports importing and exporting ODT, RTF, HTML, Markdown, and plain-text documents.

PDF export produces selectable text and supports links and bookmarks, making it useful for distributing reports and other finished documents.

The ability to work with common document formats is one of WordCraft's main advantages. However, opening a DOCX file does not guarantee that every element will render identically to Microsoft Word. Complex layouts, embedded objects, and less common formatting features should be checked after import.

Styles, Themes, and Page Layout

WordCraft includes paragraph and character styles, style sets, themes, and a live style gallery. Users can create or modify styles to maintain consistent formatting throughout a document.

Page-layout controls include margins, page size, orientation, columns, section breaks, page breaks, headers, footers, page numbers, watermarks, and page borders.

The application also supports automatic hyphenation, line numbering, drop caps, and text wrapping around pictures and shapes.

These capabilities are useful for creating structured reports, academic papers, business documents, and longer publications.

Tables, Images, and Graphics

WordCraft supports tables with merged and split cells, borders, shading, banded styles, sorting, and formulas. Rows can split across pages, and table headers can repeat on subsequent pages.

Images can be inserted, resized, cropped, rotated, recolored, and adjusted for brightness, contrast, and transparency. Additional features include picture styles, shapes, and text boxes with floating positioning and text wrapping.

The editor also supports converting text into tables and tables into text, helping users restructure information without recreating content manually.

References and Academic Writing

WordCraft provides a collection of tools for structured and academic documents, including:

• Automatic tables of contents.

• Footnotes and endnotes.

• Citations and bibliographies in APA, MLA, Chicago, and IEEE styles.

• Captions and tables of figures.

• Cross-references and indexes.

• Tables of authorities.

These tools make WordCraft particularly interesting for students, researchers, and professionals who work with long documents containing references and supporting material.

Users preparing formal academic submissions should verify that the generated references and formatting comply with their institution's requirements.

Comments, Track Changes, and Proofreading

WordCraft includes comments displayed in document margins or a dedicated pane. Track Changes allows users to review edits and accept or reject individual changes.

Document comparison, spelling and grammar suggestions, a thesaurus, word counting, and an accessibility checker provide additional support for reviewing and refining documents.

The application also includes document inspection and editing restrictions.

These features are valuable for collaborative editing and document review, although teams should test their specific review workflows before replacing established Microsoft Word processes.

Mail Merge

WordCraft supports mail merge using CSV data sources and document fields. Users can insert merge fields, configure address blocks and greeting lines, preview personalized documents, and generate merged output.

Envelope and label tools are also included.

This can help automate repetitive administrative tasks such as generating letters, notices, certificates, and personalized business documents.

AI and Command-Line Automation

WordCraft is designed to support both manual editing and programmatic document manipulation. Its editing engine exposes commands that are shared across the graphical interface, keyboard shortcuts, command search, CLI, JSON control channel, and MCP server.

The command-line interface supports document conversion, text extraction, structural inspection, rendering, and scripted editing.

For example, convert a Word document to PDF:

wordcraft-cli convert report.docx report.pdf

Extract its text:

wordcraft-cli text report.docx

Inspect the document structure as JSON:

wordcraft-cli inspect report.docx

These capabilities make WordCraft useful for batch processing, document-generation pipelines, and AI-assisted workflows. Compatible agents can inspect documents and execute editing commands without relying exclusively on screenshots or mouse interaction.

Download WordCraft v0.15.0 - Software Mirrors
WordCraft v0.15.0 for Windows wordcraft-0.15.0-windows-x86.msi | 86.02 MB wordcraft-0.15.0-windows-x86-portable.zip | 104.65 MB wordcraft-0.15.0-windows-x64.msi | 87.69 MB wordcraft-0.15.0-windows-x64-portable.zip | 108.43 MB wordcraft-0.15.0-windows-arm64.msi | 84.5 MB wordcraft-0.15.0-windows-arm64-portable.zip | 105.29 MB
WordCraft v0.15.0 for macOS wordcraft-cli-0.15.0-macos-universal.zip | 95.04 MB wordcraft-0.15.0-macos-universal.dmg | 121.21 MB
WordCraft v0.15.0 for Linux wordcraft-0.15.0-linux-x86_64.tar.gz | 108.98 MB wordcraft-0.15.0-linux-x86_64.rpm | 109.89 MB wordcraft-0.15.0-linux-x86_64.pkg.tar.zst | 82.84 MB wordcraft-0.15.0-linux-x86_64.flatpak | 77.61 MB wordcraft-0.15.0-linux-x86_64.deb | 110.5 MB wordcraft-0.15.0-linux-x86_64.AppImage.zsync | 181.08 kB wordcraft-0.15.0-linux-x86_64.AppImage | 103.33 MB wordcraft-0.15.0-linux-riscv64.tar.gz | 107.56 MB wordcraft-0.15.0-linux-riscv64.rpm | 108.46 MB wordcraft-0.15.0-linux-riscv64.deb | 108.89 MB wordcraft-0.15.0-linux-aarch64.tar.gz | 105.94 MB wordcraft-0.15.0-linux-aarch64.rpm | 107.53 MB wordcraft-0.15.0-linux-aarch64.pkg.tar.zst | 81.51 MB wordcraft-0.15.0-linux-aarch64.flatpak | 74.72 MB wordcraft-0.15.0-linux-aarch64.deb | 108.26 MB wordcraft-0.15.0-linux-aarch64.AppImage.zsync | 176.15 kB wordcraft-0.15.0-linux-aarch64.AppImage | 100.52 MB
WordCraft v0.15.0 for BSD wordcraft-0.15.0-freebsd-x86_64.tar.gz | 108.76 MB wordcraft-0.15.0-freebsd-aarch64.tar.gz | 104.9 MB
Others Download related to WordCraft v0.15.0 wordcraft-web-0.15.0.zip | 19.27 MB
WordCraft v0.15.0 Source Code WordCraft v0.15.0 Source code (zip) WordCraft v0.15.0 Source code (tar.gz)
WordCraft v0.15.0 Release Notes: What's Changed Use the body boundary for header margin anchors by @finwarman in #2485 Keep repeated spaces in LaTeX export by @Silvertree2010 in #5051 Paragraph side borders lie outside the text (#5036) by @VoiceLessQ in #5088 Lists: a level above counts its first number at the first deeper item by @pakiat in #5108 Layout: tab leader dots spaced and placed as in Word by @pakiat in #5144 Tables: a middle or bottom cell of a row split across pages lines up on the first page by @pakiat in #5145 Tables: a row isn't split to leave a picture cell empty beside its text; it moves on whole, as in Word (#4272) by @hsmith-dev in #5161 .doc import: the default font is the stylesheet's standard font, so Word 365 files open in Aptos (#5066) by @D13GOOOO in #5069 Layout tests: compare pairs of layouts drawn with the same faces by @pakiat in #5167 Columns: the line between them is drawn for selected text, as tall as the section's text (#5178) by @D13GOOOO in #5183 Shapes: arrowheads keep their kind and size and are drawn at Word's sizes by @pakiat in #5192 Markdown export: a list inside a block quote stays in the quote by @MichaelMuClaude in #5156 DOCX: keep alternate-content table rows and cells by @OPUZZLES in #5221 review.compare: keep text boxes of both documents in the result by @Silvertree2010 in #5248 Count the runs proofing text walks, instead of timing it (#5185) by @Hamza0101 in #5190 #3660: refuse a package that declares too many zip entries before zip parses them by @ljkx in #5260 Shapes: align soft-edge paint and caret opacity by @finwarman in #1600 DOCX: VML horizontal rules take their line by @Morzilla in #3956 Review: add a Word-style Proofing task pane with clickable suggestions by @totoz64 in #4471 Files: documents keep their embedded alternative format chunks (w:altChunk), and MHT opens by @naxvog in #1104 CP1.3: proofing moves to craft-proof (craft-libs) and craft-i18n language-data by @benandez in #5202 CP2.3: English spelling reads SCOWL 2026 (#40, #520, #1281, #1045) by @benandez in #5205 CP4.3–CP4.5: Language packs — proof.packs/installPack/removePack, Options › Language packs, missing-pack bar, About › Language data by @benandez in #5208 New Contributors @hsmith-dev made their first contribution in #5161 @ljkx made their first contribution in #5260 @totoz64 made their first contribution in #4471 Full Changelog: v0.14.0...v0.15.0

Performance and Compatibility

WordCraft is written in Rust and uses a modular architecture for document editing, typography, pagination, rendering, and file conversion. Its interface uses the native GPU-oriented egui framework rather than Electron or Tauri.

The project emphasizes responsive editing through cached paragraph layout and on-demand page rendering. Its developers report that typing in a 188-page document triggers layout recalculation in approximately 1.4 milliseconds. This is a project-reported benchmark rather than an independent performance test.

WordCraft supports macOS, Windows, Linux, BSD, and browsers through WebAssembly. It can read and write DOCX files and supports several additional document formats.

The main limitation is compatibility maturity. The project's roadmap identifies real-world DOCX testing, native printing, and signed application builds as remaining work. Charts, SmartArt, the equation editor, and the Draw tab are also listed as future development areas.

Consequently, WordCraft should not yet be assumed to offer complete feature parity with Microsoft Word, particularly for complex business documents and specialized workflows.

System Requirements

• Operating systems: Windows, macOS, Linux, and BSD.

• Browser support: WebAssembly build for compatible browsers.

• Document formats: DOCX, ODT, RTF, HTML, Markdown, and plain text.

• Export: PDF and page images, in addition to supported document formats.

• Build requirements: Rust and the project's documented dependencies.

The project does not provide a comprehensive minimum hardware specification in the information reviewed. Actual performance will depend on document complexity, fonts, embedded images, and system resources.

Pros and Cons

Pros

• Free and open-source under the MIT or Apache 2.0 license.

• Familiar ribbon-style interface inspired by Microsoft Word.

• DOCX reading and writing support.

• Extensive formatting, styles, tables, and page-layout tools.

• References, citations, footnotes, indexes, and tables of contents.

• Comments, Track Changes, and proofreading tools.

• Mail merge with CSV data sources.

• PDF export with selectable text, links, and bookmarks.

• Native desktop applications and a browser-based version.

• CLI and MCP support for document automation.

• Offline spelling and grammar features without requiring cloud processing.

Cons

• Still approaching maturity for everyday writing and production use.

• Complex DOCX compatibility needs more real-world validation.

• Native printing and signed application builds remain development priorities.

• Charts, SmartArt, and the equation editor are not yet complete features.

• Does not provide full Microsoft Word feature parity.

• Professional workflows may require additional testing and workarounds.

How to Install

The simplest option is to download a package from the project's GitHub Releases page, if a suitable build is available for your operating system.

To build and run WordCraft from source, install Rust and the required development dependencies, then execute:

git clone https://github.com/storytold/wordcraft
cd wordcraft
cargo run --release -p wordcraft -- --sample

To open an existing document:

cargo run --release -p wordcraft -- report.docx

For browser-based use, navigate to the web application directory and start the development server:

cd apps/wordcraft-web
trunk serve

The development server will provide a local address for opening WordCraft in a browser. Refer to the official repository for detailed build dependencies, release packages, and platform-specific instructions.

Final Verdict

WordCraft is an ambitious open-source word processor that combines familiar document editing with advanced formatting, references, review tools, mail merge, and automation. Its support for DOCX and PDF, together with additional formats and offline functionality, makes it a promising alternative for users who want more control over their document workflow.

Its command-line interface and MCP integration are especially interesting for developers and organizations that generate or process documents programmatically. The extensive feature set also makes it worth exploring for students, writers, and office users.

The biggest concern is maturity. WordCraft covers many of Microsoft Word's ribbon features, but feature coverage is not the same as complete behavioral compatibility. Users who depend on complex documents, advanced diagrams, equations, or established printing workflows should test their files carefully before switching.

For everyday writing and experimentation with open-source office software, WordCraft is worth trying. For mission-critical document production, it is better treated as a developing alternative until its compatibility and printing workflows have been validated against real-world requirements.

Free

Download

Software Informations:
Developer:
ArtCraft
Operating System:
Windows / macOS / Linux / BSD
Date Added:
2026-11-03T04:00:52.567Z