Quick Start
If you are choosing between Steam, a standalone release, and a source checkout, read Install options first.
Complete Development Setup first.
1. Launch
uv run python launcher.pyThe launcher starts the cooperating services and reports selected ports. Use the reported main URL; http://127.0.0.1:48911 is only the preferred default.
2. Configure providers
Open /api_key on the main URL. Select a Core provider and, if needed, an Assist provider. Enter credentials for the selected provider and run connectivity checks. Provider/model lists are revision-specific; do not copy old screenshots.
3. Open chat
A fresh data root is initialized from locale-specific character defaults and the bundled Yui Lolita asset. Character identifiers/display names come from active data; the old fixed “小天” claim is no longer valid.
Text chat uses the shared React component. Voice availability depends on the selected core/TTS path and microphone permission.
4. Customize and inspect
| Route | Purpose |
|---|---|
/character_card_manager | Character data and avatar selection |
/model_manager | Avatar model management |
/live2d_emotion_manager | Live2D emotion mapping |
/vrm_emotion_manager | VRM emotion mapping |
/voice_clone | Provider-dependent voice cloning |
/memory_browser | Memory outputs and processing state |
/api_key | Provider and credential settings |
Not every provider or deployment mode supports every feature. Agent and hosted-plugin capabilities require the agent/tool service.