Writing Tools: A System-Wide Grammar Assistant That Puts LLM Choice in Your Hands

What Problem It Solves and Who It Is For

The key constraint is that it is not a standalone grammar engine. It depends on an external LLM, which means the quality of output is tied to the model you pick. For users who want offline operation or who are wary of sending text to a cloud service, local LLM support is a critical feature. The README specifically mentions Ollama, llama.cpp, KoboldCPP, TabbyAPI, and vLLM as local backends, so the project is designed for privacy-conscious users who have some technical comfort. For everyone else, the free Gemini API is the quickest path, but that requires an internet connection and a Google account.

The Mechanism: Hotkey Invocation and Pop-Up Interface

The architecture is not described in detail in the material, but the repository is Swift-based, and the macOS port is a native Swift app. The Windows and Linux versions are likely built with a different framework, but that is not confirmed. What matters is that the app runs in the background, likely as a tray icon, and listens for the global hotkey. The README mentions settings are accessed from the tray icon on Windows, and you can enable starting on boot. The pop-up interface appears to be a separate window that overlays the current app, which is how it can replace text in place without switching contexts.

Getting It Running: Installation and Configuration

The installation process is deliberately simple for Windows, but the configuration of the LLM is the real setup step. The README claims a one-minute installation, which is true only if you already have an API key or a local server running. For a local LLM, you need to install and run Ollama or another server first, which adds time. The project does not bundle a model, so you must understand what an API key is or how to run a local model. This is a barrier for non-technical users, but it is also what gives the app its flexibility.

Limitations and Failure Modes

The most obvious limitation is that the app is only as good as the LLM it connects to. If the free Gemini API has rate limits or requires a paid tier for higher usage, that could affect heavy users, though the README does not mention limits. The claim that it works in any language is plausible but depends on the model's multilingual ability. Another failure mode is the reliance on a global hotkey and text replacement. Some applications may not support programmatic text replacement, or they may have their own hotkey conflicts. The README does not address these edge cases. Privacy is a stated priority, but if you use a cloud LLM, your selected text is sent to that provider. The README says your API key and config stay on your device and there is no logging by the app, but it cannot control what the LLM provider does with the text. The project is GPL-3.0, which means if you modify and distribute it, your changes must be open source under the same license. That is fine for personal use, but it may be a constraint for companies that want to embed it in proprietary software. The Linux version is explicitly a work-in-progress, so Linux users may encounter bugs or missing features. Finally, the project depends on a single maintainer who is a high school student, which raises questions about long-term sustainability, though the recent release history shows active development through 2026.

Alternatives and How They Differ

The most direct alternative is Apple Intelligence Writing Tools, but that is limited to Apple silicon Macs and certain regions. Writing Tools explicitly positions itself as a replacement for that, and it also works on Intel Macs and in the EU, which Apple does not cover. Grammarly is another alternative, but it uses rule-based NLP, not LLMs, and it is a subscription service. Writing Tools argues that LLMs are more capable of understanding context and following custom instructions, such as 'translate to French' or 'add comments to code'. Grammarly also works system-wide on some platforms, but it does not offer the same level of custom prompt control. A more technical alternative is to use a local LLM with a dedicated text editor plugin, such as Ollama with a VS Code extension, but that does not provide system-wide hotkey access. Writing Tools consolidates that into a single app. The key difference is the breadth of model support: Writing Tools can use any OpenAI-compatible API, which includes most cloud providers, plus several local servers. That flexibility is rare in a system-wide writing assistant.

Maintenance, Upgrade Cost, and License Implications

The project is under active development, with releases like Win_v9+mac_OS_v6.1 in May 2026 and earlier versions in March and December 2025. That suggests a regular release cadence, but the maintenance cost for users is moderate. Because it is a portable app on Windows, updates likely require downloading a new zip and replacing the folder, which is simple but manual. The README does not describe an auto-update mechanism. On macOS, the native Swift port is maintained by a separate contributor, so updates may be staggered across platforms. The GPL-3.0 license means you are free to use and modify the code for personal use, but if you distribute a modified version, you must make the source available under the same license. This is not legal advice, but it is a standard implication. There is no mention of a paid tier or premium features, and the README states it will always be free. The project is supported by donations via Buy Me a Coffee, which is a sustainability risk but not a direct cost to users. Before adopting it, you should verify that the latest release works on your OS version and that your chosen LLM backend is compatible.

Editorial conclusion

Writing Tools is for users who want Apple Intelligence-like writing assistance on Windows, Linux, or Intel Macs, and who are comfortable configuring an LLM backend, either a free Gemini API key or a local model. It is not for users who expect a plug-and-play grammar checker with no setup, or who need strict privacy guarantees without reading the code. Before adopting it, verify that your preferred LLM provider is supported, check the current release for your OS, and review the GPL-3.0 license if you plan to modify or redistribute it. The project is actively maintained as of mid-2026, with recent releases for both Windows and macOS, so check the latest version first.

Official sources

1. Issues
2. License: GPL-3.0
3. README
4. Releases
5. theJayTea/WritingTools on GitHub