← 返回 Skills 市场
Quant
作者
77Spongebob
· GitHub ↗
· v1.0.0
847
总下载
1
收藏
9
当前安装
1
版本数
在 OpenClaw 中安装
/install quant
功能描述
智能量化投资助手，支持多源数据获取、因子计算、多引擎回测、实时风控和交易信号推送。
安全使用建议
This skill looks like a legitimate quant helper but has several practical inconsistencies you should resolve before installing or providing credentials: (1) Ask the author or maintainer why TUSHARE_TOKEN (or config.tushare_token) is not declared in the registry metadata — do not paste your tushare token until you review the code. (2) Confirm how `quant install` / `quant setup` are implemented: there is no CLI wrapper in the manifest, so check what commands the agent will run to 'auto install' dependencies. (3) Request the missing modules referenced in SKILL.md (factors.py, backtest.py, risk.py) or a minimal reproducible install/run guide; current SKILL.md advertises capabilities not present in the package. (4) If you plan to use real trading (signal → execution), insist on explicit, auditable confirmation steps for any real-money operations. If you cannot verify these points, treat the skill as untrusted and avoid supplying API tokens or running automatic install commands.
功能分析
Type: OpenClaw Skill
Name: quant
Version: 1.0.0
The skill provides quantitative trading functionality, including data retrieval from Tushare/Akshare and factor analysis in lib/data.py and lib/alpha_stream.py. It is classified as suspicious because it utilizes high-risk capabilities such as network access and file system interaction, and the SKILL.md file contains prompt-injection instructions designed to autonomously direct the agent's workflow ('I will immediately create...'), which attempts to bypass user control. Additionally, the mention of an automated installation command ('quant install') and discrepancies between the documented directory structure and the provided files suggest potential for unverified code execution.
能力评估
Purpose & Capability
The name/description (quantitative investment assistant) aligns with the included Python modules (data access and factor/alpha code). However the SKILL.md references additional modules (factors.py, backtest.py, risk.py) and CLI commands (quant setup, quant install, quant data, etc.) that are not present in the file manifest or registry metadata. That mismatch (advertised functionality vs. provided files) reduces confidence that the skill will behave as described.
Instruction Scope
SKILL.md instructs the agent to run CLI commands such as `quant setup` and `quant install` and promises to 'immediately create lib/data.py and config.yaml skeleton'. In this package the data.py and config.yaml already exist, but there is no provided CLI binary or wrapper in the manifest. The instructions also assert 'all data processed locally, no exfiltration' — there is no code enforcing this (the code fetches remote data via tushare/akshare/yfinance). The instructions are therefore vague and grant broad discretion to the agent (e.g., to install dependencies or create files) without a clear, reproducible runtime plan.
Install Mechanism
There is no install specification (instruction-only skill). That limits automatic installation risk, but SKILL.md tells the agent it will 'auto install dependencies' on `quant install` despite no install steps being declared. If the agent runs pip/apt/brew commands at runtime, it will perform network installs — a normal behavior for such a skill but one the user should be aware of since the install commands are not specified or reviewable in the manifest.
Credentials
The registry metadata declares no required environment variables, but lib/data.py reads os.getenv('TUSHARE_TOKEN') when attempting to call tushare.pro_api. config.yaml also contains a tushare_token field. This is a mismatch: the skill expects (or will behave differently with) a secret token but does not declare it in requires.env/primaryEnv. No other unrelated credentials are requested, but the missing declaration and the token dependency are noteworthy because the user may be prompted to supply credentials later.
Persistence & Privilege
always is false, there are no config paths requested, and the code does not attempt to modify other skills or system-wide agent settings. The skill does mention creating files and installing dependencies, but that is normal for a code-providing skill and is contained to its own files.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install quant - 安装完成后，直接呼叫该 Skill 的名称或使用
/quant触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
quant 1.0.0 首发上线！
- 提供A股及全球市场的数据获取、因子挖掘、回测、风控与实时信号推送功能
- 支持 tushare、akshare、yfinance 数据接口
- 集成多回测引擎，便捷策略开发与复盘
- 内置50+主流及另类量化因子
- 严格本地数据安全及下单操作确认
- 用户友好命令行快速上手与配置指导
元数据
常见问题
Quant 是什么？
智能量化投资助手，支持多源数据获取、因子计算、多引擎回测、实时风控和交易信号推送。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 847 次。
如何安装 Quant？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install quant」即可一键安装，无需额外配置。
Quant 是免费的吗？
是的，Quant 完全免费（开源免费），可自由下载、安装和使用。
Quant 支持哪些平台？
Quant 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Quant？
由 77Spongebob（@77spongebob）开发并维护，当前版本 v1.0.0。
推荐 Skills