knowledge-card-factory

智能知识卡片生产流水线。自动化完成热点发现、内容深挖、AI 配图、多端发布的完整流程。适用于自媒体运营、知识分享、品牌营销等场景。

Security Scan

The skill's declared purpose (automated discovery → research → image generation → multi‑platform publish) matches its instructions and there are no hidden installs, env vars, or odd endpoints; it is internally coherent.

Purpose & Capability

Name/description (content production pipeline) align with the runtime instructions and declared required skills (brave-search, agent-reach, nano-banana-pro, xiaohongshu-mcp). All required capabilities are relevant to producing and publishing knowledge cards; nothing unrelated (e.g., cloud admin credentials or unrelated tooling) is requested.

Instruction Scope

SKILL.md instructs cross‑platform content aggregation (agent-reach) and automated publishing. This stays within the skill's purpose, but the instructions enable broad scraping/aggregation across multiple social platforms and extraction of images/data. That behavior is expected for the stated workflow but is broad in scope and may have privacy/copyright and platform policy implications. The skill itself does not request additional host files/envs, nor does it instruct reading system paths outside its outputs.

Install Mechanism

Instruction-only skill with no install spec and no code files executed locally—lowest install risk. The README suggests installing dependent skills via clawhub, which is expected and uses the platform's package mechanism rather than arbitrary downloads.

Credentials

The skill declares no required env vars or credentials itself, which is appropriate. However it depends on other skills (agent-reach, xiaohongshu-mcp, nano-banana-pro) that likely require platform credentials or API keys to fetch or publish content. The user should inspect those dependent skills for any credential requests before enabling the combo.

Persistence & Privilege

always is false, the skill is user‑invocable and may be invoked autonomously (platform default). It does not request permanent system presence or modify other skills/config; error handling mentions local drafts/outputs which is within its scope.

Assessment

This combo appears coherent with its stated purpose. Before installing or using it: (1) Review the dependent skills (brave-search, agent-reach, nano-banana-pro, xiaohongshu-mcp) to see what credentials or scopes they require and where they send data; (2) confirm you have permission to aggregate/scrape the target platforms and to republish their content (copyright/privacy); (3) keep require_confirmation enabled (default) to avoid accidental publishing; (4) test in a sandbox account first to verify outputs and notifications (feishu/email) and to understand what metadata and source links are stored in output/sources.json; (5) if you need stricter limits, restrict platforms or disable automated publish steps.

智能知识卡片生产流水线 — 从选题到发布的全自动化解决方案。

何时使用

当用户需要：

• 快速生产社交媒体内容
• 制作知识卡片/信息图
• 自动化内容发布流程
• 热点追踪与借势营销

触发关键词：

• "做一张知识卡片"
• "帮我生成内容发小红书"
• "自动化内容生产"
• "热点内容创作"

工作流程

1. 热点发现 (brave-search)

# 使用 brave-search 搜索热点
# 示例: 搜索 AI 行业趋势
search_query = "{用户指定的主题} 最新 趋势 2026"

输出:

• 热点话题列表
• 相关新闻报道
• 关键数据点

2. 内容深挖 (agent-reach)

# 使用 agent-reach 跨平台抓取
# 支持: Twitter/X, 小红书, B站, 公众号, 微博, LinkedIn

操作:

1. 基于热点话题搜索多平台内容
2. 提取核心观点和数据
3. 生成内容摘要

3. 卡片创作

方案 A: AI 配图 (nano-banana-pro)

提示词模板:
"Create a {style} style illustration about {topic}, 
featuring {key_elements}, modern, clean design"

方案 B: 卡片渲染 (card-renderer)

支持风格:
- Mac Pro 风格
- 赛博朋克
- 包豪斯
- 清新简约

4. 多端发布 (xiaohongshu-mcp)

发布配置:
  platform: xiaohongshu
  content:
    title: {生成的标题}
    body: {生成的正文}
    images: [{图片路径}]
    tags: [{话题标签}]

使用示例

示例 1: 制作 AI 趋势卡片

用户指令:

"帮我做一张 AI Agent 发展趋势的知识卡片，发到小红书"

执行步骤:

# Step 1: 热点发现
topics = brave_search("AI Agent 发展趋势 2026")

# Step 2: 内容深挖
content = agent_reach(
    platforms=["twitter", "xiaohongshu", "wechat"],
    query="AI Agent trends"
)

# Step 3: 生成配图
image = nano_banana_pro(
    prompt="AI Agent ecosystem diagram, futuristic style",
    size="1024x1024"
)

# Step 4: 发布
result = xiaohongshu_publish(
    title="2026 AI Agent 五大趋势",
    content=content.summary,
    images=[image]
)

输出:

✅ 知识卡片已发布
📄 标题: 2026 AI Agent 五大趋势
🔗 链接: https://xiaohongshu.com/note/xxx
👀 预览: [卡片图片]

示例 2: 天气出行指南

用户指令:

"帮我做一张北京周末出行天气指南"

执行步骤:

# Step 1: 获取天气
weather_data = weather("北京", days=3)

# Step 2: 搜索热门景点
spots = brave_search("北京周末热门景点")

# Step 3: 渲染卡片
card = card_renderer(
    template="清新简约",
    data={
        weather: weather_data,
        recommendations: spots
    }
)

配置选项

workflow.json

{
  "name": "knowledge-card-factory",
  "version": "1.0.0",
  "stages": [
    {
      "id": "discover",
      "skills": ["brave-search"],
      "config": {
        "result_limit": 10,
        "freshness": "week"
      }
    },
    {
      "id": "research",
      "skills": ["agent-reach"],
      "config": {
        "platforms": ["twitter", "xiaohongshu", "wechat"],
        "max_results": 20
      }
    },
    {
      "id": "create",
      "skills": ["nano-banana-pro", "card-renderer"],
      "config": {
        "default_style": "cyberpunk",
        "image_size": "1024x1024"
      }
    },
    {
      "id": "publish",
      "skills": ["xiaohongshu-mcp"],
      "config": {
        "auto_publish": false,
        "require_confirmation": true
      }
    }
  ],
  "error_handling": {
    "retry": 3,
    "on_failure": "save_draft"
  }
}

错误处理

错误类型	处理方式
搜索失败	重试 3 次，使用缓存数据
配图生成失败	降级到文字卡片
发布失败	保存本地草稿，通知用户

注意事项

1. 发布前确认: 默认开启 require_confirmation，确保用户审核后再发布
2. 内容质量: 生成的内容需要用户确认，避免 AI 幻觉
3. 图片版权: AI 生成图片需标注来源
4. 平台规则: 不同平台有不同发布限制，注意合规

扩展能力

• 添加新渠道: 实现对应平台的 Skill 接口
• 自定义模板: 在 templates/ 目录添加卡片模板
• 数据源扩展: 在 sources/ 目录添加新数据源