Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

投教工作流

v1.0.0

[何时使用]当用户需要投资者教育内容时;当用户说'写篇投教文章'、'解释这个投资概念'、'做个投教卡片'、'定投是什么'、'如何防骗'时触发。基于 LLM Wiki 机制的投教内容生产与分发工作流。

0· 22·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for lj22503/investor-education-workflow.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "投教工作流" (lj22503/investor-education-workflow) from ClawHub.
Skill page: https://clawhub.ai/lj22503/investor-education-workflow
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Canonical install target

openclaw skills install lj22503/investor-education-workflow

ClawHub CLI

Package manager switcher

npx clawhub@latest install investor-education-workflow
Security Scan
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Benign
high confidence
Purpose & Capability
Name/description describe an investor-education content workflow and the SKILL.md only asks for knowledge-retrieval, targeted search, plain-language conversion, multi-format output, and feedback into a wiki — all coherent with that purpose. Declared related skills (ljg-*, searxng, url-to-markdown, task-state-tracker) match the described pipeline.
Instruction Scope
Instructions are scoped to knowledge-base retrieval, targeted web search, content conversion and writing back to the wiki. It explicitly allows Exec, Read, Write and WebSearch tools; the SKILL.md itself does not instruct reading arbitrary system files, but the allowed toolset would permit broader actions if the runtime agent is given those capabilities. The workflow also encourages storing user/interaction records in the wiki/raw layers — review how customer data (especially PII) is handled and retained.
Install Mechanism
No install spec and no code files — instruction-only skill is lowest install risk. Nothing is downloaded or extracted; no third-party packages are pulled by this skill itself.
Credentials
The skill declares no required environment variables, credentials, or config paths. That aligns with its instruction-only nature. Note: it relies on other skills and external search/output services — those may require credentials or network access, so verify their credential needs separately.
Persistence & Privilege
always:false and no install behavior; the skill does not request permanent elevated presence. It instructs writing back into the LLM Wiki (its own knowledge store), which is expected for a knowledge workflow.
Assessment
This skill appears coherent and low-risk as provided, but before installing: 1) Verify the trustworthiness of the related skills it calls (ljg-*, searxng, url-to-markdown, task-state-tracker) because those handle web scraping, conversion, storage, or external APIs. 2) Confirm runtime permissions: Exec/Read/Write/WebSearch are broad — ensure the agent environment restricts Exec to safe commands and prevents arbitrary system access. 3) Data policy: the workflow writes user interactions into a wiki/raw store — decide and document how PII and client-specific data are filtered, anonymized, and retained. 4) Validate external search sources and url-to-markdown behavior to avoid pulling untrusted content or introducing supply-chain risk. If those points are acceptable, the skill is consistent with its stated purpose.

Like a lobster shell, security has layers — review code before you run it.

behavioral-financevk97fh73gvm6hqqfq44t7qt3f6x85fjjbcontent-productionvk97fh73gvm6hqqfq44t7qt3f6x85fjjbinvestor-educationvk97fh73gvm6hqqfq44t7qt3f6x85fjjbknowledge-basevk97fh73gvm6hqqfq44t7qt3f6x85fjjblatestvk97fh73gvm6hqqfq44t7qt3f6x85fjjbllm-wikivk97fh73gvm6hqqfq44t7qt3f6x85fjjb
22downloads
0stars
1versions
Updated 6h ago
v1.0.0
MIT-0

investor-education-workflow: 投教工作流 🎯

📋 功能描述

帮助用户系统化生产与分发投资者教育内容。基于 LLM Wiki 机制,实现"知识库优先 → 搜索补充 → 转译大白话 → 多形式输出 → 反馈入库"的完整闭环。

适用场景:

  • 投教内容生产(文章/卡片/语音/视频脚本)
  • 知识点转译(专业术语 → 大白话)
  • 客户问答响应(1 对 1/群发/讲座)
  • 知识库维护(LLM Wiki编译与更新)

边界条件:

  • 不替代持牌投教机构
  • 输出为 Markdown/HTML/PNG,需配合 ljg-skills 实现多形式输出
  • 知识库优先检索,缺失时定向搜索官方来源

🔄 核心处理流程(5 阶段)

触发需求 → ① 检索知识库 → ② 若缺失,定向搜索 → ③ 转译大白话 → ④ 多形式输出 → ⑤ 反馈入库
阶段动作调用 Skill输出
① 检索知识库在五大模块中全文检索ljg-learn(概念解剖)官方素材/经典案例/标准话术
② 定向搜索知识库无内容/内容过时/问题具体searxng + url-to-markdown原始材料(标注来源与时间)
③ 转译大白话去术语化/场景化/行为化/可视化ljg-plain(白话说)≤200 字素人版解释 + 顾问口播脚本
④ 多形式输出按场景分发ljg-card(可视化)+ ljg-writes(深度文章)文字/图文/语音/视频/卡片
⑤ 反馈入库结构化回填至知识库task-state-tracker使用频次/理解难度/迭代标记

📚 五大模块知识库

模块内容对应 Skill
模块 1:投资基本功金融基础/基本面分析/技术分析/财经素养ljg-learn, fund-analyzer-pro
模块 2:市场认知与风险扫描资本市场全景/风险识别/合规交易decision-checklist, content-compliance
模块 3:行为管理行为金融学/常见偏误/策略与矫正decision-checklist, ljg-relationship
模块 4:投资规划与生命周期财务规划/资产配置/退休规划fund-allocator, ljg-rank
模块 5:投资者陪伴与持续成长分层学习路径/多元形式/日常陪伴companion-script, ljg-plain, ljg-card

详细五大模块说明 → references/five-modules.md LLM Wiki 架构 → references/llm-wiki.md 四专家思维框架 → references/four-experts.md


⚠️ 常见错误

错误 1:直接给专业解释,不转译

问题:
• 用户问"什么是市盈率",输出 PE=股价/每股收益
• 客户听不懂,教育失败

解决:
✓ 必须使用 ljg-plain 转译成大白话
✓ 用生活类比("PE=回本年限,像买店铺看租金")
✓ 输出≤200 字,结构:是什么→举例→注意点

错误 2:忽略知识库优先原则

问题:
• 每次都重新搜索,不查知识库
• 内容质量不稳定,可能引用非官方来源

解决:
✓ 优先检索五大模块知识库
✓ 命中→直接转译;未命中→定向搜索官方来源
✓ 标注来源与时间,确保合规

错误 3:单一形式输出

问题:
• 只输出 Markdown 文字,不考虑使用场景
• 客户在微信/讲座/短视频等不同场景无法使用

解决:
✓ 按场景选择输出形式(文字/图文/语音/视频/卡片)
✓ 使用 ljg-card 生成可视化素材
✓ 输出多形式版本,供顾问选择

🧪 使用示例

输入:

客户问:"什么是市盈率(PE)?能不能说人话?"

预期输出:

  • 阶段①:检索知识库 → 找到官方"一图看懂市盈率"
  • 阶段③:转译大白话 → "PE 就是你买这家公司,靠利润需要多少年回本..."
  • 阶段④:多形式输出 → 文字版(200 字)+ 图文版(对比表)+ 语音条(30 秒)
  • 阶段⑤:反馈入库 → 标记"客户接受度高",关联"行为偏误 - 锚定效应"

输入:

写篇投教文章:定投为什么能对抗择时焦虑?

预期输出:

  • 识别场景:投教内容生产
  • 调用:ljg-learn(概念解剖)→ ljg-plain(白话说)→ ljg-writes(深度文章)→ ljg-card(可视化)
  • 输出:Markdown 文章 + PNG 卡片

🔧 故障排查

问题检查项
不触发description 是否包含触发词?用户输入是否匹配场景?
知识库无内容是否优先检索五大模块?搜索源是否正确(中国投资者网/交易所)?
转译不通俗是否调用 ljg-plain?是否使用生活类比?字数是否≤200?
输出形式单一是否调用 ljg-card?是否按场景选择形式?
未反馈入库是否调用 task-state-tracker?是否标记使用频次/理解难度?

🔗 相关资源

  • 五大模块知识库:references/five-modules.md
  • LLM Wiki 架构:references/llm-wiki.md
  • 四专家思维框架:references/four-experts.md
  • 报告模板:templates/education-template.md
  • 标准参考:docs/SKILL-STANDARD-v3.md

Comments

Loading comments...