Back to Details

wind-find-finance-skill

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

触发时机

Trigger Conditions

用户问以下任一情况时触发:
  1. "有什么金融能力 / 推荐什么金融工具 / 平台能做什么"
  2. 提了具体金融问题(行情 / 基金 / 财务 / 估值 / 选股 / 回测 / 复盘 / 公告 / 宏观等)但 AI 不确定用哪个 skill
  3. 问 AIMarket / Wind 平台元问题
Trigger when the user asks any of the following:
  1. "What financial capabilities do you have? / What financial tools do you recommend? / What can the platform do?"
  2. Raises specific financial questions (market quotes / funds / finance / valuation / stock selection / backtesting / review / announcements / macroeconomics, etc.) but the AI is unsure which skill to use
  3. Asks meta questions about the AIMarket / Wind platform

不触发场景

Non-trigger Scenarios

用户已明确指定某个具体 skill("用 wind-mcp-skill 查茅台" / "用 buffett 分析这家公司")→ 直接走那个 skill,不绕本入口。
If the user has clearly specified a specific skill (e.g., "Use wind-mcp-skill to check Moutai" / "Use buffett to analyze this company"), directly route to that skill and do not use this entry.

推荐策略(按用户提问类型路由)

Recommendation Strategy (Route by User Question Type)

用户提问类型推什么推几个
取数 / 查询(行情、基金、财务、公告、新闻、宏观)数据 skill(目前 = wind-mcp-skill)1 个
做分析(估值 / 复盘 / 选股 / 回测 / 个股研究 / 市场主线)数据 skill + 对应分析 skill 组合2 个
探索("你们能做啥" / "我想研究 A 股")各 category 各 1 个样例3-5 个
永远附 wind-mcp-skill 作数据底座,除非用户明确不要数据。
User Question TypeWhat to RecommendNumber of Recommendations
Data Retrieval / Query (market quotes, funds, finance, announcements, news, macroeconomics)Data skill (currently = wind-mcp-skill)1
Analysis (valuation / review / stock selection / backtesting / individual stock research / market trends)Combination of data skill + corresponding analysis skill2
Exploration ("What can you do?" / "I want to research A-shares")1 sample from each category3-5
Always include wind-mcp-skill as the data base, unless the user explicitly states they don't need data.

流程

Process

  1. 用 Read 读 
    references/skills-catalog.md
    → 拿全清单
  2. 判别用户提问类型(取数 / 分析 / 探索)
  3. 按推荐策略筛 1-5 个相关 skill
  4. 每个推荐 skill 输出装包卡片(见模板)
  5. 末尾固定追加升级提示一句话(见下)
  1. Use Read to load 
    references/skills-catalog.md
    → Get the full list
  2. Determine the user's question type (data retrieval / analysis / exploration)
  3. Filter 1-5 relevant skills according to the recommendation strategy
  4. Output an installation card (see template) for each recommended skill
  5. Append a one-sentence upgrade prompt (see below) at the end

装包卡片模板

Installation Card Template

每个推荐的 skill 按以下格式给:
推荐 <name> · <一句话描述>
为什么:<基于用户问题的一句话解释>

安装命令(国外 GitHub):
  npx skills add JsonCodeChina/wind-skills --skill <name> -g -y

安装命令(国内 Gitee 镜像):
  npx skills add https://gitee.com/jsonCodeChina/wind-skills.git --skill <name> -g -y

[如果 catalog "装好需配置" 列 = "API Key",追加这一段:]
首次使用提示:装好后向我提一个金融数据问题,我会引导你登录
aimarket.wind.com.cn 拿 API Key

[如果是分析 skill 且没在同次推荐里附数据 skill,追加:]
配套数据:推荐同时装 wind-mcp-skill 作数据底座
Format each recommended skill as follows:
Recommend <name> · <one-sentence description>
Why: <one-sentence explanation based on the user's question>

Installation command (GitHub overseas):
  npx skills add JsonCodeChina/wind-skills --skill <name> -g -y

Installation command (Gitee mirror domestic):
  npx skills add https://gitee.com/jsonCodeChina/wind-skills.git --skill <name> -g -y

[If the "Post-installation Configuration" column in the catalog = "API Key", append this section:]
First-use prompt: After installation, ask me a financial data question, and I will guide you to log in to aimarket.wind.com.cn to get an API Key

[If it's an analysis skill and the data skill is not included in the same recommendation, append:]
Supporting data: It is recommended to install wind-mcp-skill as the data base at the same time

参数说明(必读)

Parameter Explanation (Must Read)

  • -g
    / 
    --global
    :装到 
    ~/.agents/skills/
    (用户级,跨项目共享,金融机构内网友好)
  • -y
    / 
    --yes
    :必加,跳过交互菜单(不加会卡)
  • --skill <name>
    :从 monorepo 抽指定子 skill 装;不写会装全部 skill
  • -g
    / 
    --global
    : Install to 
    ~/.agents/skills/
    (user-level, shared across projects, friendly for internal networks of financial institutions)
  • -y
    / 
    --yes
    : Mandatory, skips the interactive menu (will get stuck if not added)
  • --skill <name>
    : Extract and install the specified sub-skill from the monorepo; if not specified, all skills will be installed

升级提示(每次回答末尾固定追加)

Upgrade Prompt (Append at the end of every response)

> 想看最新平台能力可跑 `npx skills update -g -y` 同步本地清单。
含义:
update
重拉所有已装 skill 最新版,
-g
只升级全局 skill,
-y
跳过 scope 提示。
> To see the latest platform capabilities, run `npx skills update -g -y` to sync the local list.
Meaning: 
update
pulls the latest versions of all installed skills, 
-g
only upgrades global skills, 
-y
skips the scope prompt.

边界

Boundaries

  • 本 skill 不调用任何 MCP server,不需要 API Key
  • 本 skill 不写用户本地任何文件
  • 不做远端 WebFetch diff(catalog.md 由 
    npx skills update -g -y
    自动同步,不需要 AI 实时比对)
  • references/skills-catalog.md
    是 skill 包打包时的快照,跟着 skill 包一起 push / update
  • This skill does not call any MCP server, no API Key required
  • This skill does not write any files to the user's local machine
  • Do not perform remote WebFetch diff (catalog.md is automatically synced by 
    npx skills update -g -y
    , no real-time comparison by AI needed)
  • references/skills-catalog.md
    is a snapshot taken when the skill package is built, and is pushed/updated along with the skill package