CSCB/vibe-mouse

What it solves

Vibe-mouse provides a customizable human-computer interaction (HCI) layer that streamlines the use of AI-powered IDEs (like Cursor, Trae, and Windsurf). It eliminates the need to memorize complex keyboard shortcuts by mapping AI actions—such as inline editing, chatting, and accepting code—to a wide variety of of external hardware devices (mice, gamepads, voice, etc.).

How it works

The project uses a pluggable adapter architecture to capture events from various input devices and route them through a DeviceManager to an Executor that triggers the corresponding system shortcuts. It includes a "Skill System" that allows users to define complete work modes (e.g., "Code Review" or "Debug Helper") which can override shortcuts, inject specific system prompts into LLMs, and change hardware feedback settings.

For voice interaction, it bridges speech recognition to LLMs (via Huawei Cloud Token Plan/MaaS) and can optionally use a "Prompt Refine" step to convert colloquial speech into structured prompts before the AI generates code.

Who it’s for

Developers who use AI IDEs and want to optimize their workflow through physical hardware triggers, voice commands, or automated context-switching between different coding tools.

Highlights

• Multi-Device Support: Pluggable adapters for mice, keyboards, gamepads, Bluetooth, IR remotes, HID devices, and IoT/Network triggers.
• AI IDE Integration: Built-in shortcut mappings for Trae, Cursor, Windsurf, VS Code (Copilot), DevEco Studio, and CodeArts.
• Auto Window Detection: Automatically detects which AI IDE is in focus and switches the active shortcut configuration accordingly.
• Skill System: Configurable modes that combine system prompts, specific shortcut overrides, and hardware feedback (LEDs, vibration, sound).
• Voice-to-LLM Bridge: Direct integration with LLMs to generate code via voice, with an optional prompt optimization layer.