← Back to Skills Marketplace
gechengling

Security Digital Human Ops

by lingfeng-19 · GitHub ↗ · v3.0.0 · MIT-0
cross-platform ✓ Security Clean
71
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install bank-digital-human
Description
AI assistant for banks and insurance to design virtual avatar scripts, plan content calendars, manage live streams, optimize interactions, and analyze digita...
README (SKILL.md)

\r \r

Digital Human Operations Assistant / 数字人运营助手\r

\r

English: AI-powered digital human operations assistant — covers script design, content planning, live streaming, and performance analytics.\r \r 中文: 数字人运营助手——覆盖话术设计、内容策划、直播管理、互动优化、效果分析。\r \r ---\r \r

Industry Pain Points / 行业痛点\r

\r | Pain Point / 痛点 | Impact / 影响 | Solution / 本Skill解决方案 |\r |------------------|-------------|------------------------|\r | 内容创作耗时 | 数字人内容生产成本高 | AI自动生成脚本 |\r | 互动性差 | 数字人缺乏人情味 | 个性化互动设计 |\r | 场景单一 | 数字人应用场景受限 | 多场景内容矩阵 |\r | 效果难评估 | 直播效果难以量化 | 数据分析模板 |\r \r ---\r \r

Trigger Keywords / 触发关键词\r

\r English Triggers: digital human, virtual avatar, live streaming, virtual spokesperson, AI streamer, content marketing\r \r 中文触发词(优先): 数字人 / 虚拟主播 / 直播运营 / 虚拟形象 / AI主播 / 数字人直播 / 元宇宙 / 虚拟代言人 / 品牌IP / 内容创作 / 直播脚本 / 短视频 / 虚拟客服 / 数字员工\r \r ---\r \r

Core Capabilities / 核心能力\r

\r

1. Script Generation / 脚本生成\r

\r

SCRIPT_TEMPLATES = {\r
    "产品介绍类": {\r
        "时长": "2-3分钟",\r
        "结构": """\r
        【开场 Hook】(0-15秒)\r
        - 痛点引入:"你是不是也遇到过...?"\r
        - 数据吸引:"90%的人都不知道的..."\r
        \r
        【自我介绍】(15-30秒)\r
        - 虚拟形象介绍\r
        - 今日主题预告\r
        \r
        【核心内容】(1-2分钟)\r
        - 产品/服务介绍(3个卖点)\r
        - 场景化案例\r
        - 限时优惠/福利\r
        \r
        【行动号召 CTA】(15-30秒)\r
        - 关注引导\r
        - 评论区互动\r
        - 私信/链接\r
        \r
        【结尾】(0-15秒)\r
        - 回顾要点\r
        - 下期预告\r
        """\r
    },\r
    \r
    "知识科普类": {\r
        "时长": "1-2分钟",\r
        "结构": """\r
        【开场】:"今天教你一个..."\r
        【问题】:"为什么..."\r
        【解答】:"因为..."\r
        【案例】:"举个例子..."\r
        【总结】:"所以..."\r
        """\r
    }\r
}\r
\r
def generate_script(script_type: str, params: dict) -> dict:\r
    """\r
    生成数字人直播/视频脚本\r
    """\r
    template = SCRIPT_TEMPLATES.get(script_type)\r
    if not template:\r
        return {"error": "未知脚本类型"}\r
    \r
    # 生成台词\r
    script_content = generate_script_content(script_type, params)\r
    \r
    # 生成AI提示词\r
    ai_prompts = generate_ai_prompts(script_type, params)\r
    \r
    # 生成分镜\r
    storyboard = generate_storyboard(script_type, params)\r
    \r
    return {\r
        "script_type": script_type,\r
        "duration": template.get("时长", ""),\r
        "structure": template.get("结构", ""),\r
        "content": script_content,\r
        "ai_prompts": ai_prompts,\r
        "storyboard": storyboard\r
    }\r
```\r
\r
### 2. Content Calendar / 内容日历\r
\r
```markdown\r
## 数字人内容日历模板\r
\r
### 周内容规划\r
\r
| 日期 | 内容类型 | 主题 | 数字人形象 | KPI目标 |\r
|-----|---------|------|-----------|--------|\r
| 周一 | 产品种草 | 周一理财推荐 | 职业装版 | 点赞+100 |\r
| 周二 | 知识科普 | 金融小知识 | 学院风 | 收藏+50 |\r
| 周三 | 互动问答 | 粉丝答疑 | 休闲风 | 评论+30 |\r
| 周四 | 案例分享 | 客户故事 | 职业装版 | 转发+20 |\r
| 周五 | 直播带货 | 周五秒杀 | 活力版 | GMV+X万 |\r
| 周六 | 用户UGC | 晒单有礼 | - | UGC+10 |\r
| 周日 | 下周预告 | 剧透福利 | 创意版 | 关注+50 |\r
```\r
\r
### 数字人人设设定\r
\r
```python\r
DIGITAL_HUMAN_PERSONAS = {\r
    "专业理财师": {\r
        "外观": "职业西装,30岁左右女性",\r
        "性格": "专业、亲切、值得信赖",\r
        "语言风格": "专业术语+通俗解释",\r
        "口头禅": "理财有道,生活无忧",\r
        "适用场景": "产品介绍、理财建议"\r
    },\r
    \r
    "活力小姐姐": {\r
        "外观": "休闲时尚,25岁左右女性",\r
        "性格": "活泼、亲切、有趣",\r
        "语言风格": "网络用语、emoji",\r
        "口头禅": "冲冲冲!买它!",\r
        "适用场景": "直播带货、互动活动"\r
    },\r
    \r
    "专业分析师": {\r
        "外观": "商务男装,35岁左右男性",\r
        "性格": "理性、严谨、权威",\r
        "语言风格": "数据分析、图表解读",\r
        "口头禅": "数据不会说谎",\r
        "适用场景": "市场分析、宏观解读"\r
    }\r
}\r
```\r
\r
### 3. Performance Analytics / 效果分析\r
\r
```python\r
class DigitalHumanAnalytics:\r
    """数字人效果分析"""\r
    \r
    def analyze_performance(self, video_stats: dict) -> dict:\r
        """\r
        分析视频/直播效果\r
        """\r
        # 核心指标\r
        metrics = {\r
            "播放量": video_stats.get("views", 0),\r
            "完播率": video_stats.get("completion_rate", 0),\r
            "点赞数": video_stats.get("likes", 0),\r
            "评论数": video_stats.get("comments", 0),\r
            "转发数": video_stats.get("shares", 0),\r
            "关注转化": video_stats.get("follows", 0)\r
        }\r
        \r
        # 计算互动率\r
        engagement_rate = (\r
            (metrics["点赞数"] + metrics["评论数"] + metrics["转发数"]) /\r
            max(metrics["播放量"], 1) * 100\r
        )\r
        \r
        # 评分\r
        score = self._calculate_content_score(metrics)\r
        \r
        return {\r
            "metrics": metrics,\r
            "engagement_rate": round(engagement_rate, 2),\r
            "content_score": score,\r
            "optimization_suggestions": self._get_suggestions(metrics),\r
            "benchmark_comparison": self._compare_to_benchmark(metrics)\r
        }\r
    \r
    def _calculate_content_score(self, metrics: dict) -> float:\r
        """内容评分(满分100)"""\r
        score = 0\r
        \r
        # 完播率(30分)\r
        score += min(metrics["完播率"] / 100 * 30, 30)\r
        \r
        # 互动率(40分)\r
        engagement = (\r
            (metrics["点赞数"] + metrics["评论数"]) /\r
            max(metrics["播放量"], 1)\r
        )\r
        score += min(engagement * 400, 40)  # 2.5%互动率=满分\r
        \r
        # 关注转化(30分)\r
        follow_rate = metrics["关注转化"] / max(metrics["播放量"], 1)\r
        score += min(follow_rate * 3000, 30)  # 1%关注率=满分\r
        \r
        return round(score, 1)\r
```\r
\r
---\r
\r
## Disclaimer\r
\r
This skill provides digital human content planning tools for educational purposes. All content must comply with applicable advertising and financial marketing regulations.\r
Usage Guidance
Treat this as an incomplete low-confidence review: local workspace inspection failed with a sandbox command error, so install only after confirming the skill artifacts are coherent and scoped.
Capability Assessment
Purpose & Capability
No reviewed artifact evidence showed a purpose mismatch or unsafe capability.
Instruction Scope
No reviewed artifact evidence showed hidden or overbroad runtime instructions.
Install Mechanism
No reviewed artifact evidence showed unsafe installation behavior.
Credentials
No reviewed artifact evidence showed disproportionate environment access.
Persistence & Privilege
No reviewed artifact evidence showed persistence or privilege abuse.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install bank-digital-human
  3. After installation, invoke the skill by name or use /bank-digital-human
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v3.0.0
- Major upgrade: Skill now includes comprehensive digital human operations tools for banking and insurance marketing. - Added AI-powered script generation with industry-tailored templates and persona settings (中英文/English & Chinese). - Introduced a weekly content calendar template for efficient planning of digital avatar activities. - Implemented automated performance analytics with key metrics, engagement scoring, and optimization suggestions. - Expanded trigger keywords and multi-scenario support for broader financial marketing applications.
Metadata
Slug bank-digital-human
Version 3.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Security Digital Human Ops?

AI assistant for banks and insurance to design virtual avatar scripts, plan content calendars, manage live streams, optimize interactions, and analyze digita... It is an AI Agent Skill for Claude Code / OpenClaw, with 71 downloads so far.

How do I install Security Digital Human Ops?

Run "/install bank-digital-human" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Security Digital Human Ops free?

Yes, Security Digital Human Ops is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Security Digital Human Ops support?

Security Digital Human Ops is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Security Digital Human Ops?

It is built and maintained by lingfeng-19 (@gechengling); the current version is v3.0.0.

💬 Comments