Description

港股 IPO 打新研究助手。抓取实时数据（孖展、基石、评级、暗盘、A+H折价、中签率），供 AI 分析判断。触发词：港股打新、新股分析、IPO、孖展、保荐人、暗盘、中签率、基石投资者。不适用：A 股打新、美股 IPO、基金申购。

README (SKILL.md)

港股打新研究助手

Name: Hk Ipo Research Assistant
Author: marvae

数据来自第三方公开网站，不构成投资建议。

安装依赖

首次使用前安装 Python 依赖：

cd \x3Cskill_dir>
pip install -r scripts/requirements.txt

需要 Python 3.10+。

调用方式

所有命令通过 Python 调用：

cd \x3Cskill_dir>
python3 scripts/hkipo.py \x3C命令> [参数]

以下文档中的命令（如 overview、analyze 02692）都省略了前缀，实际调用时加上 python3 scripts/hkipo.py。

快速开始

cd \x3Cskill_dir>

# 当前招股一览
python3 scripts/hkipo.py overview

# 一键分析（聚合多维度数据）
python3 scripts/hkipo.py analyze 02692

# 单只股票分项查询
python3 scripts/hkipo.py aipo ipo-brief 02692        # 基本信息
python3 scripts/hkipo.py aipo margin-detail 02692    # 孖展明细
python3 scripts/hkipo.py aipo cornerstone 02692      # 基石投资者
python3 scripts/hkipo.py aipo rating-detail 02692    # 机构评级

完整命令清单

概览类

命令	说明
`overview`	当前招股 IPO 一览（格式化输出）
`calendar`	资金日历，按截止日期分组
`analyze \x3C代码>`	一键分析（聚合基本面+孖展+基石+评级+保荐人历史）
`profile`	用户画像 + 当前 IPO 数据（读取用户画像，输出当前 IPO 数据供 AI 分析）

AiPO 数据（主力数据源）

命令	说明
`aipo margin-list`	孖展列表（当前招股）
`aipo margin-detail \x3C代码>`	单只孖展明细（13+券商）
`aipo rating-list`	新股评级列表
`aipo rating-detail \x3C代码>`	单只评级明细（各机构打分）
`aipo ipo-brief \x3C代码>`	IPO 基本信息（保荐人、发行价、市值、PE）
`aipo cornerstone \x3C代码>`	基石投资者（名单、金额、锁定期、解禁日）
`aipo placing \x3C代码>`	配售结果

暗盘数据

命令	说明
`aipo grey-list`	暗盘历史列表
`aipo grey-today`	今日暗盘
`aipo grey-trades \x3C代码> --date YYYY-MM-DD`	暗盘成交明细
`aipo grey-prices \x3C代码> --date YYYY-MM-DD`	暗盘分价表
`aipo grey-placing \x3C代码>`	暗盘配售详情
`aipo allotment`	中签结果列表
`aipo scroll`	市场滚动消息

排名统计

命令	说明
`aipo bookrunner-rank --start-date --end-date`	账簿管理人排名
`aipo broker-rank --start-date --end-date`	券商参与度排名
`aipo stableprice-rank --start-date --end-date`	稳价人排名
`aipo summary --year 2025`	年度 IPO 统计
`aipo performance --year 2025`	年度表现统计
`aipo by-office --year 2025`	按注册地统计

历史数据（集思录）

命令	说明
`jisilu list`	历史 IPO 列表（表格）
`jisilu list --json`	历史 IPO 列表（JSON）
`jisilu list --sponsor XX`	按保荐人筛选
`jisilu detail \x3C代码>`	单只历史详情

估值计算

命令	说明
`ah compare \x3C代码> --price \x3C发行价> --name \x3C名称>`	A+H 折价计算

中签率预测

命令	说明
`odds --oversub \x3C倍数> --price \x3C价格>`	中签率表格（甲/乙组各档）
`allotment --oversub \x3C倍数> --price \x3C价格>`	单次中签率预测

其他数据源

命令	说明
`hkex active`	港交所活跃申请（含招股书链接）
`tradesmart list`	TradeSmart 数据
`futu list`	富途历史数据

市场情绪判断

命令	说明
`hsi`	恒生指数当日表现（大盘情绪）
`sponsor`	保荐人历史排名（胜率、平均首日表现）
`sentiment sponsor-search --name \x3C名称>`	查特定保荐人战绩
`etnet list`	保荐人排名（经济通数据，备用）
`etnet search --name \x3C名称>`	查保荐人（经济通数据）

结合以下数据综合判断：

大盘情绪：HSI 涨跌反映整体风险偏好
保荐人战绩：胜率 >80% 的保荐人历史表现更好
孖展金额：>50亿热门，>100亿爆款
数据源 Fallback：AASTOCKS 挂了自动切换 etnet

输出格式

默认 JSON：aipo、jisilu --json、ah compare
格式化文本：overview、calendar、odds、jisilu list
加 --format table：aipo 命令可切换表格输出

分析要点

详见 references/analysis-guide.md，核心：

市场热度：孖展 >50亿热门，>100亿爆款
基石投资者：数量 + 质量 + 锁定期
保荐人：用 jisilu list --sponsor 查历史战绩
估值：PE 对比同行，A+H折价 >30% 有安全边际
中签率：
1. 从 margin-detail 获取 oversubscription_forecast（预测超购倍数）
2. 用 ./hkipo odds --oversub \x3C倍数> --price \x3C发行价> 生成表格
3. 把表格给用户，让用户根据本金自己决定打几手

用户画像

使用 profile 命令：

运行 ./hkipo profile
如果没有配置，输出会告诉你需要问用户哪些问题
问用户：本金、风险偏好、是否用孖展、券商
把答案写入 scripts/config/user-profile.yaml
再次运行 profile 获取数据

配置文件 (scripts/config/user-profile.yaml)：

capital: 22000           # 港币
risk: conservative       # conservative/balanced/aggressive
margin: never            # never/cautious/active
broker: longbridge

详细参考

references/analysis-guide.md — 分析框架详解
references/ipo-mechanism.md — 机制详解（回拨、红鞋、绿鞋）
references/aipo-api.md — AiPO API 完整文档

Usage Guidance

This skill appears coherent and implements what it claims: it scrapes public IPO data sources and provides CLI analysis. Before installing, take these precautions: (1) inspect scripts/requirements.txt and run pip inside an isolated virtualenv or container; (2) review the network endpoints the tool will call (aipo.myiqdii.com, aastocks.com, hkex, sinajs/qt.gtimg, futu, tradesmart, etc.) to ensure you are comfortable with outbound connections; (3) be aware the tool will write local cache and a user profile under scripts/ (scripts/config/user-profile.yaml); (4) if you use private credentials for any broker APIs later, check whether the code ever adds new environment variable requirements — currently none are requested. If you want extra assurance, run the CLI in a read-only or sandboxed environment and audit scripts/cache.py and scripts/requirements.txt before use.

Capability Analysis

Type: OpenClaw Skill Name: hk-ipo-research-assistant Version: 0.1.1 The skill is classified as suspicious primarily due to a potential prompt injection vulnerability against the AI agent, as described in `SKILL.md`. The instructions explicitly direct the agent to ask the user for sensitive financial information (capital, risk preference, broker) and then to "把答案写入 `scripts/config/user-profile.yaml`" (write the answers to `scripts/config/user-profile.yaml`). If the AI agent writes raw, unsanitized user input directly into this YAML configuration file, it could lead to YAML injection, potentially allowing a malicious user to compromise the agent or the local environment when the file is later processed. While the Python code uses `yaml.safe_load` for reading, this mitigates only some, but not all, forms of YAML injection. Additionally, the local storage of sensitive user financial data (capital, broker) in `scripts/config/user-profile.yaml` is a risk, even if not immediately exfiltrated. The extensive network access to various financial data sources is expected for the skill's stated purpose and does not show signs of malicious intent.

Capability Assessment

✓ Purpose & Capability

Name/description match the implementation: the package contains adapters and CLI commands for fetching IPO-related data (AiPO, AAStocks, HKEX, Futu, TradeSmart, Jisilu), analytics (allotment prediction, A+H comparison), and helper references. The requested actions (pip install, run python scripts/hkipo.py) are proportionate to the stated purpose.

✓ Instruction Scope

SKILL.md directs installing Python deps and running the bundled CLI. Runtime behavior (HTTP requests to third‑party public data sites, parsing HTML/JSON, generating predictions, reading/writing scripts/config/user-profile.yaml, caching) is within scope. The skill reads/writes only files inside its directory (scripts/config/, cache files) and does not instruct reading arbitrary system files or environment variables. It does request the user to enter a user profile (capital, risk, margin, broker) which is saved locally.

ℹ Install Mechanism

No platform package manager install spec; SKILL.md requires 'pip install -r scripts/requirements.txt'. Installing Python dependencies is expected for a Python CLI but introduces normal supply-chain risk (pip packages). The requirements.txt file is included in the bundle; users should inspect it before pip install or use an isolated virtualenv. No direct downloads from untrusted URLs were seen in the skill files themselves.

✓ Credentials

The skill declares no required environment variables, no credentials, and no external config paths. All network calls target public data providers needed for market data (see note). It does not request AWS/GitHub/other unrelated keys.

✓ Persistence & Privilege

always:false and default invocation behavior. The skill persists local caches and a user profile under scripts/, which is reasonable for this tool. It does not modify other skills or system-wide agent settings.

Version History

v0.1.1

**Python 脚本化并增加依赖说明：** - 明确要求 Python 3.10+，需先安装依赖（pip install -r ...）。 - 所有命令统一用 python3 scripts/hkipo.py 调用，替代原有 bash 脚本。 - 文档中的命令调用方式更新，增加依赖安装与入口说明。 - 其他功能、命令清单、输出格式等保持不变。

v0.1.0

港股 IPO 打新研究助手首次发布，支持多维数据采集与分析。 - 提供实时港股 IPO 数据（孖展、基石、评级、暗盘、中签率等）统一查询 - 支持一键分析、单股分项查询、基金资料、券商及保荐人排名、历史数据 - 包含用户画像配置与定制化分析流程 - 多数据源自动切换，支持表格及 JSON 格式输出 - 提供估值计算、中签率预测、市场情绪和赛道分析等辅助决策功能

v1.1.0

Add sentiment module: HSI index + sponsor historical performance

v1.0.2

Initial release

v1.0.1

Remove user config from package

v1.0.0

Initial release

Metadata

Slug hk-ipo-research-assistant

Version 0.1.1

License —

All-time Installs 7

Active Installs 7

Total Versions 6

Frequently Asked Questions

What is Hk Ipo Research Assistant?

港股 IPO 打新研究助手。抓取实时数据（孖展、基石、评级、暗盘、A+H折价、中签率），供 AI 分析判断。触发词：港股打新、新股分析、IPO、孖展、保荐人、暗盘、中签率、基石投资者。不适用：A 股打新、美股 IPO、基金申购。 It is an AI Agent Skill for Claude Code / OpenClaw, with 787 downloads so far.

How do I install Hk Ipo Research Assistant?

Run "/install hk-ipo-research-assistant" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Hk Ipo Research Assistant free?

Yes, Hk Ipo Research Assistant is completely free (open-source). You can download, install and use it at no cost.

Which platforms does Hk Ipo Research Assistant support?

Hk Ipo Research Assistant is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Hk Ipo Research Assistant?

It is built and maintained by m3 (@marvae); the current version is v0.1.1.

More Skills

Hk Ipo Research Assistant