菜单
Controls the 3D AI therapist avatar's real-time emotional responses, mirroring, active listening behaviors, and empathic reactions. Creates the feeling of talking to a real, present human therapist.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Ahmet-Talha-Kavakli/Lyra-DB --skill emotion-mirroring-avatar命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | emotion-mirroring-avatar |
| description | Controls the 3D AI therapist avatar's real-time emotional responses, mirroring, active listening behaviors, and empathic reactions. Creates the feeling of talking to a real, present human therapist. |
The 3D avatar is the therapist's physical presence. This skill defines how it listens, responds, mirrors, and creates genuine human connection. The goal: the user forgets they're talking to an AI.
Real therapists communicate constantly through:
The avatar must do all of this in real-time.
interface ListeningBehaviors {
microNods: boolean; // Small nods every 3-5 seconds
eyeContact: boolean; // Maintain realistic (not robotic) eye contact
breathingVisible: boolean; // Subtle chest movement
microExpressions: boolean; // Subtle emotional responses as user speaks
headTilt: number; // Slight tilt = interest (5-10 degrees)
blinkRate: number; // Natural blink rate (12-15 per minute)
}
The avatar mirrors the user's emotional state with a slight lag and at a reduced intensity:
interface MirroringRule {
userEmotion: EmotionLabel;
avatarResponse: AvatarExpression;
intensity: number; // 0-1, always lower than user's
lagMs: number; // Realistic mirror delay: 200-800ms
}
const mirroringRules: MirroringRule[] = [
{
userEmotion: 'sadness',
avatarResponse: 'soft_empathy',
intensity: 0.4, // Subdued — not matching full sadness
lagMs: 400
},
{
userEmotion: 'joy',
avatarResponse: 'warm_smile',
intensity: 0.6, // Match joy but not exuberantly
lagMs: 200
},
{
userEmotion: 'anger',
avatarResponse: 'calm_concern', // DO NOT mirror anger — de-escalate
intensity: 0.3,
lagMs: 600
},
{
userEmotion: 'fear',
avatarResponse: 'warm_steady', // Steady, calm — anchor for scared user
intensity: 0.2,
lagMs:
}
];
Critical rule: NEVER mirror anger. Mirror calm and concern instead. An angry mirror escalates conflict.
When therapist says something like "That sounds really hard":
When user has an insight or "aha" moment:
When user reveals something difficult:
When therapist is formulating a response:
interface LipSyncConfig {
blendShapeMode: 'realtime' | 'precomputed';
phonemeModel: 'allosaurus' | 'wav2vec' | 'whisper_phonemes';
smoothingFactor: number; // 0.1-0.3 for natural movement
languageCode: string; // 'en-US', 'tr-TR', etc.
}
Silence is a tool. Avatar handles silence intentionally:
| Silence Duration | Avatar Behavior |
|---|---|
| 0-5 seconds | Maintain eye contact, subtle nod |
| 5-15 seconds | Remain present, no pressure |
| 15-30 seconds | Slight lean forward, optional soft check-in |
| 30+ seconds | Gently invite: "Take your time. I'm here." |
Never fill silence anxiously — this mirrors user's discomfort.
// Session engine sends events → Avatar system responds
type AvatarEvent =
| { type: 'SPEAKING_START'; text: string; audioBuffer: ArrayBuffer }
| { type: 'SPEAKING_END' }
| { type: 'EMOTION_RESPOND'; emotion: EmotionLabel; intensity: number }
| { type: 'MIRROR_USER'; userEmotion: EmotionLabel }
| { type: 'LISTENING_MODE' }
| { type: 'THINKING_PAUSE' };