Skill flagged — suspicious patterns detected

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

A Stock Daily Market Sense

v1.6.2

A Stock Daily Market Sense,用于基于 Tushare Pro A 股 daily 日线数据生成盘后市场研报。当用户要求每日盘面感知、盘后复盘、赚钱效应在哪里、上涨主线是什么、主线 vs 资金轮动、爆量下跌、低位放量异动、指数背离个股、该弱不弱就是强、历史某日复盘,或基于 daily/da...

0· 52·0 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 chinfi-codex/a-stock-daily-market-sense.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "A Stock Daily Market Sense" (chinfi-codex/a-stock-daily-market-sense) from ClawHub.
Skill page: https://clawhub.ai/chinfi-codex/a-stock-daily-market-sense
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

Bare skill slug

openclaw skills install a-stock-daily-market-sense

ClawHub CLI

Package manager switcher

npx clawhub@latest install a-stock-daily-market-sense
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
Name, description, SKILL.md, template, and scripts/market_panel.py all align: the skill fetches Tushare A-share daily data, computes volume/price features, and produces a market evidence pack used to render the six-section report. The code and template are consistent with the declared purpose.
Instruction Scope
SKILL.md instructs running scripts/market_panel.py to fetch Tushare data and build candidate pools; the script reads TUSHARE_TOKEN from the environment or a cwd .env file. The runtime instructions do not instruct reading unrelated system files or contacting unexpected third-party endpoints beyond Tushare. Note: the SKILL.md explicitly allows reading a .env file in the working directory, which can expose any secrets stored there.
Install Mechanism
This is an instruction-only skill with no install spec; there is no package download or privileged installer. That minimizes install-time risk. However, the Python script requires runtime dependencies (pandas, tushare) which are not declared in registry metadata.
!
Credentials
Registry metadata lists no required env vars, but both SKILL.md and the script require a TUSHARE_TOKEN (read from environment or cwd/.env). Dependencies (pandas, tushare) are required at runtime but not declared. Asking for a single Tushare API token is appropriate for this purpose, but the metadata omission and .env file access are inconsistent and worth noting.
Persistence & Privilege
always is false and the skill does not request persistent system-wide privileges. The script reads cwd/.env and environment variables and makes network calls to Tushare (expected). It does not declare modifying other skills or agent config. Autonomous invocation is allowed by default but not combined with other high-risk flags.
What to consider before installing
This skill appears to do what it claims: it fetches A-share daily data from Tushare and builds the numeric evidence the report uses. Before installing, note these points: (1) You must provide a TUSHARE_TOKEN (either as an environment variable or in a .env file in the skill's working directory); the registry metadata did not declare this — verify you are comfortable exposing that token to the agent and store a token with minimal privileges if possible. (2) The Python script requires pandas and tushare at runtime but the skill metadata doesn't list dependencies — install them in a controlled virtualenv/sandbox. (3) The script reads cwd/.env; avoid putting any unrelated secrets in that file or run the skill in an isolated folder. (4) The script makes network calls to Tushare (expected). If you need extra assurance, inspect the full scripts/market_panel.py (we reviewed the visible portions) or run it in an isolated environment to confirm it only contacts Tushare and does not write unexpected files or contact 3rd-party endpoints. If you cannot verify these things, treat the metadata mismatch as a red flag and consider running it manually rather than enabling autonomous invocation.

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

latestvk9786nnsdmwwse0nsndvrsc86985gs1m
52downloads
0stars
1versions
Updated 2d ago
v1.6.2
MIT-0

A Stock Daily Market Sense

目标与边界

基于 Tushare 历史交易数据,生成 A 股盘后市场感知报告,回答四个问题:

  1. 盘面温度如何:普涨、普跌、局部主线、情绪退潮,还是指数掩盖结构分化。
  2. 赚钱效应在哪里:哪些股票真的涨了,且成交额足以证明资金参与。
  3. 风险在哪里:哪些股票在爆量下跌,可能对应高位筹码松动、主线分歧、事件冲击或弱势出清。
  4. 结构性机会在哪里:哪些个股出现低位放量异动;指数弱时哪些股票“该弱不弱就是强”。

不用于单股深度基本面、港股/美股/基金/期货/加密资产、分钟级实时交易、自动下单、组合优化或任何买卖建议。

核心原则

成交额优先。 强弱判断必须用成交额支撑,不能只看涨跌幅。涨跌幅是结果,成交额变化才是资金行为痕迹。

主题主线由模型临时归纳,不套现成行业/概念口径。先读候选池,再根据业务事实、产品环节、产业链位置、事件催化、量价同步性分组。少于 3 只股票或共同性不足的方向,归入“其他/孤立强势股”。

不要把单日上涨说成趋势。持续性至少看 3-5 个交易日;不确定时直接写不确定,不硬凑主线。

固定数据流程

  1. 确定交易日 D:解析用户日期;若给的是非交易日,用 trade_cal 取 D 日及以前最近交易日。默认只用 <= D 数据;只有用户明确要求后验时才允许 D+N。
  2. 生成证据包:运行 scripts/market_panel.py panel,拿到盘面温度、成交额集中度、赚钱效应候选池、爆量下跌候选池、低位放量异动候选池、抗跌股候选池。
  3. 判断盘面状态:优先看涨跌家数、涨跌幅中位数、大涨/大跌数量、上涨/下跌成交额占比、总成交额、主要指数表现。
  4. 归纳赚钱效应与主线:读 money_effect_samples,先看候选总数和合计成交额,再分组,最后按 ★★★/★★/★ 评级。
  5. 识别爆量下跌风险:读 volume_decline_samples,归纳风险类型,列高强度样本;若 ★★★ 主线代表股进入爆量池,必须写风险传导提示。
  6. 识别低位异动:读 low_position_volume_anomaly_samples,按 starter / sustain / quiet / undetermined 四类呈现。
  7. 识别抗跌股:读 resilient_against_index_samples。只有指数触发弱环境时才列候选;指数不弱时直接说明本模块无输出。
  8. 按模板成稿:读取 reference/report_template.md,按六个固定模块填空,数字优先放表格,结论短句化。

数据获取

依赖:Python 3.9+、tusharepandas。敏感信息只从环境变量或当前工作目录 .env 读取:

TUSHARE_TOKEN=your_token

基础命令:

cd C:\Users\chenh\OneDrive\skills\a-stock-daily-market-sense
python scripts/market_panel.py panel --asof 20260424 --lookback 120 --index 000300.SH

查看完整参数:

python scripts/market_panel.py panel --help

常用阈值默认值:

模块核心筛选规则
赚钱效应pct_chg >= 7% 且成交额 >= 2 亿元,按成交额降序
爆量下跌pct_chg <= -3%、20 日放量倍数 >= 2.0x、成交额 >= 1 亿元
低位异动A 轨底部区域或 B 轨深回撤走平,且触发日 15 日放量倍数 >= 3.0x、涨幅 >= 7%
抗跌股仅在沪深 300 等基准指数 5 日 <= -2% 或 10 日 <= -3% 时输出;个股 5 日相对超额 >= 5pct、绝对收益 >= 0、成交额 >= 1 亿元

降级规则:

  • daily_basic 不可用:仍分析涨跌、成交额、相对强弱,但标注缺少换手率、量比、市值。
  • limit_list_d 不可用:使用 pct_chg >= 9.8% / pct_chg <= -9.8% 的近似涨跌停口径并标注。
  • 指数数据不可用:第 6 模块跳过并说明无法判断指数弱环境。
  • 候选池为空或过少:不要硬凑结论,直接说明“无清晰赚钱效应”或“无大面积爆量下跌”。

主线判定

赚钱效应候选池只负责“哪些票有资金参与”,主题和主线必须由模型基于业务事实归纳。

每个主题至少写:

  • 主题名
  • 代表股,按组内成交额取前 3-5 只
  • 共同业务事实
  • 入选数
  • 组合计成交额
  • 占赚钱效应总成交额比

主线确认度严格按以下规则:

评级判定规则
★★★ 主线组成交额占赚钱效应总成交额比 >= 30%,且 5 日涨幅中位 > 0,且 5 日相对指数超额中位 > 0,且连续放量天数中位 >= 2
★★ 潜在主线上述四条满足 3 条
★ 局部异动/资金轮动满足不超过 2 条,或候选股票数少于 3 只

若全市场上涨占比低,即便有 ★★★ 主线,也要写成“局部主线行情”,不要写成全面扩散。

风险归纳

爆量下跌不是跌幅榜复述。优先按 decline_intensity = 放量倍数 × |跌幅| 看异常程度,再按共同事实归纳风险类型。

常见风险类型:

  • 高位抱团瓦解:前期涨幅大、距 120 日高点回撤浅、当日放量崩塌。
  • 主线内部分歧:与第 3 模块强方向存在交叉,但出现高成交额放量杀跌。
  • 业绩雷/事件冲击:单日重挫且成交额跳升明显。
  • 退潮补跌:5 日已累计下跌,今日继续放量。
  • ST/风险股出清:名称含 ST/退或风险标签明显。
  • 流动性杀跌:无清晰业务叙事但放量下跌。

风险定性必须具体,例如“主线内部分歧加剧”“高位筹码松动”“弱势股出清”,不要写“加速恶化/趋稳/局部”这类空泛词。

低位异动分类

使用新版四类术语,不使用旧版“启动型/确认型/分歧型”:

类型含义
starter今日触发,后续行为尚未验证
sustain触发后成交额维持在触发日的 70% 以上,资金仍在换手
quiet触发后最近 3 日成交额缩至触发日 50% 以下,且价格未明显破位
undetermined触发过,但既未维持高换手,也未清晰缩量企稳

解读时把 sustain 和 quiet 放在重点位置;undetermined 数量高时,写“触发后资金分歧大”,不要解读成新主线。

输出规范

固定模块顺序:

  1. 盘面温度
  2. 成交额集中度与拥挤度
  3. 赚钱效应与上涨主线
  4. 亏钱效应(爆量下跌)
  5. 低位放量异动
  6. 抗跌股(该弱不弱就是强)

写作纪律:

  • 写清楚数据日期、窗口和基准指数。
  • 所有强弱判断都给成交额数字支撑。
  • 基准指数只保留关键涨跌幅,不堆数据。
  • 数字优先表格,表格后文字只解释结论。
  • 第 3 模块必须给“主线行情”或“资金轮动”明确定性。
  • 第 4 模块必须互查 ★★★ 主线代表股是否进入爆量下跌池。
  • 第 6 模块指数不弱时直接写“今日指数环境不弱,本模块无输出”,并附指数 5 日/10 日涨跌幅和门槛。
  • 不写“次日观察点”,不写买入/卖出/持有。

示例

用户输入:

使用 a stock daily market sense 生成 2026-04-24 的完整盘后报告

执行:

python scripts/market_panel.py panel --asof 20260424 --lookback 120 --index 000300.SH

输出:按 reference/report_template.md 的六模块结构生成 Markdown 研报,并在末尾标注数据口径、接口降级和“不构成投资建议”。

Comments

Loading comments...