Quick Start

This page explains how to install dependencies and run/build Live2D Copilot in a local environment.

Environment requirements

⚠️ package.json does not declare engines; the following versions are assumptions for a typical dev environment. Use the actual runnable environment as the source of truth (requires manual review).

• Node.js: An LTS version is recommended (18 or higher).
• Package manager: pnpm (the repository uses pnpm-lock.yaml).
• Rust toolchain: Rust must be installed (required by Tauri 2; a stable toolchain is recommended).
• Platform: Target platform is desktop. Currently, some of the Rust-side capabilities (mouse passthrough, gaze tracking, window screenshot) are mainly implemented on Windows (#[cfg(windows)]), so cross-platform support is limited.
• Tauri system dependencies: Install the build dependencies for your platform per the Tauri 2 official documentation.

Install dependencies

Run in the repository root:

pnpm install

Development run

In dev mode, tauri dev first starts the Vite frontend via beforeDevCommand, then brings up the Tauri window:

pnpm tauri dev

Note: In package.json, the dev script is vite (frontend only), and the tauri script is tauri. For full desktop development, pnpm tauri dev is recommended.

Build

pnpm tauri build

Before building, pnpm build (i.e., vue-tsc --noEmit && vite build) is run via beforeBuildCommand for type checking and frontend bundling.

Frontend build (standalone)

pnpm build

Key configuration

• Tauri configuration: src-tauri/tauri.conf.json (product name Live2D Copilot, identifier com.ai-zen.live2d-copilot; the window is created dynamically on the Rust side).
• Permissions declaration: src-tauri/capabilities/default.json (core:default, opener:default).
• Default model: public/models/Mao/ ("虹色Mao").
• TTS voice list: public/voices.json (available Edge-TTS voices, read and cached by the frontend).

Configuration before use

On first use, you need to configure the AI service in the Settings window:

1. Choose the model provider (DeepSeek / ChatGLM) in the Settings window.
2. Fill in the corresponding API Key.
3. Select the LLM model.
4. Select a TTS voice (default zh-CN-XiaoyiNeural).
5. Click "Save Settings".

The settings are written to settings.json under the app config directory via the Tauri Command (save_setting), and are immediately synced to the desktop pet window through the settings-updated event.

Related pages

• Product Overview: Product positioning and core features.
• Features & Architecture: Tech stack and data flow.
• Development Progress: Current version and roadmap.