ayangweb/BongoCat

If you want a small, visible companion on your desktop rather than another background utility, BongoCat is a low-commitment download. It is a desktop pet for Windows and macOS whose Live2D cat follows your cursor and reacts to keys, mouse buttons and gamepad presses; swapping in your own model is the main customization path.

The problem

Most desktop-pet projects leave you assembling a runtime or editing config before anything appears. The intended workflow here is simpler: install a release, place the cat window where it does not obstruct work, and get immediate visual feedback from the input you already produce. The step that changes is the idle screen itself — instead of a static desktop you see a cat reacting to typing and clicks. The result is a decorative layer that runs alongside your normal applications.

What's different

Events are the center of the design, not an afterthought: the README states that the right motion is matched to every key, mouse button and gamepad button, and the cat's eyes and paws follow the mouse. That input-driven animation is what separates it from a static wallpaper or a pet that merely wanders around. The window can be moved, and the project is built as a Tauri application with Rust on the backend, though that matters mainly if you intend to build from source.

Customization is intentionally narrow. You can bring your own Live2D model, and the companion Awesome-BongoCat repository collects models to download or contribute. That is a useful ecosystem if you want a different look, but the README does not present a plugin system or a deep settings menu. The other deliberate trade-off is privacy-oriented packaging: the README says no user data is collected and the app works offline with no network access. Those are author claims rather than independent audits, but they set expectations for a small, self-contained app.

The project was inspired by Bongo-Cat-Mver by MMmmmoko, and the README acknowledges that lineage. This is a desktop-toy project, not a platform for building interactive characters, so evaluate it as a finished consumer app rather than as a framework.

Who should try it

Good fit: a Windows or macOS user who wants a lightweight visual companion for a stream, a shared screen or personal desk setup, and who is content with the included model or willing to find a Live2D model from the model gallery. It is also a reasonable fit for someone who wants to watch input events reflected on screen without granting the app network access, since offline operation is documented.

Poor fit: Linux users, because the README lists only macOS and Windows as supported platforms despite the repository carrying cross-platform and Linux topics. Also a poor fit for anyone who wants a general Live2D runtime, scripting hooks or extensive per-application behavior; the documented surface is models, window placement and input reactions, not an automation platform. If you need more models, check the separate gallery before installing, because the core app is not described as shipping a large built-in library.

Requirements and limitations

You need a supported desktop OS — macOS or Windows — and a release build or a source build. The README's installation pointer is the releases page; building from source is a separate path and requires the Tauri/Rust toolchain, which is a real hurdle for casual users. The app is documented as offline and data-free in operation, so no account or network dependency is implied, but that also means model discovery happens outside the app through the model gallery.

Ongoing upkeep is mostly the usual desktop-app routine: update from releases and keep your model files available. Custom models are Live2D assets rather than arbitrary images, so sourcing and managing them is the main customization work. The README does not document a plugin ecosystem, troubleshooting matrix or automated update mechanism; if those matter, check the release notes and contributing guide before relying on the project long term.

HotRepo's take

HotRepo tracked the project once, on 2026-10-25, at 24,177 stars with +39 that day, on the daily board. That single dated observation cannot establish whether interest is sustained or fading, and no earlier board history is available. The practical decision is small: confirm that the supported-platform list matches your machine and that a model you like exists in the gallery, then treat the offline claim as the author's statement rather than an independent guarantee.