← 返回 Skills 市场
227
总下载
2
收藏
0
当前安装
11
版本数
在 OpenClaw 中安装
/install stock-trading-agents-light
功能描述
多智能体股票交易信号分析框架。当用户提供股票代码要求分析投资建议时触发。输入一只股票代码，通过7个SubAgent协作分析（基本面研究员、市场信息研究员、新闻研究员、社交媒体研究员、看涨分析师、看跌分析师、投资组合经理），输出买入/卖出/持有建议及决策依据。触发词："分析这只股票"、"给我投资建议"、"这只票值得...
安全使用建议
This skill appears to be what it says: a multi-agent stock research framework that uses Tushare and optional web search and writes a markdown report locally. Before installing or running it: 1) Keep your TUSHARE_TOKEN and BRAVE_API_KEY only in environment variables (do not paste them into chat or reports). 2) Inspect and test the redaction behavior: the skill saves full subagent outputs into a final report, so if any subagent output contains credential-like strings the redaction must reliably remove them — validate with non-production tokens first. 3) Run in an isolated environment or sandbox (virtualenv/container) because it will pip-install packages and make network calls to api.tushare.pro (and web search if configured). 4) Verify the pip install step (tushare, pandas, numpy) matches your platform policy and is acceptable. 5) Understand that the tool fetches external data (Tushare, web search) — using it transmits queries to those services. 6) For privacy/regulatory reasons, avoid feeding any personal or secret information into prompts/subagents. If you need higher assurance, review the SKILL.md and included scripts line-by-line or run them with a dummy Tushare token to observe actual behavior.
功能分析
Type: OpenClaw Skill
Name: trading-agents-simplified
Version: 2.4.2
The 'trading-agents-simplified' skill bundle is a legitimate multi-agent framework for stock market analysis. The Python scripts (get_fundamentals.py, get_market_data.py) use standard financial libraries (tushare, pandas) to fetch data from authorized APIs, and the markdown instructions (SKILL.md and references/) include extensive, explicit security rules to prevent the leakage of API tokens (TUSHARE_TOKEN, BRAVE_API_KEY) in reports or inter-agent communications.
能力评估
Purpose & Capability
The skill claims to run a multi-agent trading analysis and legitimately requires a Tushare Pro token for market/fundamental data; an optional Brave Search API key for web search is reasonable. The provided Python scripts call Tushare and compute indicators consistent with the described functionality.
Instruction Scope
Runtime instructions restrict credential access to TUSHARE_TOKEN and BRAVE_API_KEY and require redaction of any credentials from subagent outputs. The skill writes complete reports (including all subagent outputs and debate history) to ~/.openclaw/workspace/memory/reports/*.md — this is expected for the stated purpose but creates a privacy surface: if any subagent input inadvertently contains sensitive strings, the flow depends on correct redaction rules. Review and test the redaction logic before using with any sensitive inputs.
Install Mechanism
The software uses standard PyPI packages (tushare, pandas, numpy) which is proportionate. One minor inconsistency: the top metadata stated 'No install spec' while skill.json and SKILL.md include pip install instructions; this is not harmful but should be noted. No downloads from unknown hosts or archive extracts are present.
Credentials
Only TUSHARE_TOKEN is required (BRAVE_API_KEY optional). No unrelated credentials or system paths are requested. The included scripts read TUSHARE_TOKEN from the environment and use it for Tushare API calls — proportional to the stated purpose.
Persistence & Privilege
The skill is not always-enabled and does not request elevated platform privileges. It writes reports to a localized workspace path (documented). Autonomous invocation is allowed by default (platform behavior) but is not combined with any elevated credential requests or 'always: true' — no excessive persistence privileges observed.
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 stock-trading-agents-light - 安装完成后，直接呼叫该 Skill 的名称或使用
/stock-trading-agents-light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
v2.4.2
No changes detected in this version.
- Synchronized SKILL.md Output Example with research-manager.md Output Format (full Part I-V structure)
- All functionality, dependencies, and workflow remain as in the previous version.
v2.4.1
trading-agents-simplified 2.4.1
- Enhanced the research-manager to produce more complete Markdown reports.
- Skill functionality and configuration remain the same as the previous release.
v2.4.0
**Enforces stricter security and credential handling.**
- Added explicit security rules: only read credentials from environment variables, never from files.
- SubAgents and reports are prohibited from including API keys, passwords, or secrets.
- Updated documentation for secure environment variable usage; removed alternative config file loading.
- Clarified report content must redact any sensitive information.
- No changes to core workflow or output, just improved security practices and clearer instructions.
v2.3.0
- Removed PDF export and markdown-to-PDF conversion functionality.
- Dropped dependencies: reportlab and markdown Python packages.
- Updated documentation to reflect the removal of PDF export; output is now markdown-only.
- No changes to the multi-agent analysis workflow or core features.
v1.0.7
- Added pip dependencies (reportlab, markdown) to support PDF report conversion.
- Updated environment variable and dependency configuration for improved security and clarity; strongly recommends secure OpenClaw JSON config (tokens should not be stored in shell rc files).
- Clarified "side effects" and external connectivity for transparency (API usage, file writes).
- No functional changes to skill logic. These updates improve setup guidance and enhance user data security practices.
v1.0.6
- Added markdown report export: The system now outputs a full analysis report in markdown format after generating a final investment decision.
- Added PDF report generation: Introduced a script (`scripts/markdown_to_pdf.py`) to convert the markdown report into a PDF, enabling easier sharing and archiving.
- Updated documentation: SKILL.md now includes sections detailing Steps 7 and 8 for markdown export and PDF conversion, with usage instructions and requirements.
- No changes to the trading agent debate workflow or analysis logic.
v1.0.5
**Major update: Introduces a two-round debate mechanism between bull and bear researchers for improved investment decision-making.**
- Added a new "Layer 2.5" to the workflow featuring a two-round sequential debate between Bull and Bear Researchers, where each refutes and counter-refutes opposing arguments.
- Updated workflow to a 4-layer architecture: Information Gathering → Opinion Formation → Two-Round Debate → Final Decision.
- Clarified roles and tasks for Bull and Bear Researchers to include debate responsibilities with conversational, data-backed exchanges.
- Specified that the Research Manager now synthesizes all reports and the complete debate history before making the final BUY/SELL/HOLD recommendation.
- Revised the skill description to highlight the collaborative debate mechanism for richer investment analysis.
v1.0.4
- English translation of documentation and all instructions, making the entire SKILL.md fully accessible to non-Chinese users.
- No changes to code or underlying logic; documentation content and structure remain the same, but now in English.
- Maintains clear workflow steps and SubAgent descriptions, now with English naming and explanations.
v1.0.2
- Added MIT license and clarified licensing details.
- Introduced explicit environment variable requirements: TUSHARE_TOKEN (required), BRAVE_API_KEY (optional).
- Added environment and Python dependency setup instructions.
- No changes to core workflow or agent structure; documentation improvements only.
v1.0.1
- change display name
v1.0.0
Initial public release with major structural changes for the trading-agents skill.
- Rebuilt the skill around 7 SubAgent workflow: Fundamental Analyst, Market Analyst, News Analyst, Social Media Analyst, Bull Researcher, Bear Researcher, and Research Manager.
- Input is now a single stock code; system outputs buy/sell/hold recommendation plus rationale.
- Introduced references and prompts for each SubAgent (see references/*.md files).
- Provided scripts for fundamental and market data retrieval.
- Simplified overall structure: removed previous multi-layer agent/team code; focused on a leaner, parallel SubAgent execution pipeline.
- Updated documentation and usage instructions for new workflow, triggers, and output formats.
元数据
常见问题
Trading Agents 简化版 是什么？
多智能体股票交易信号分析框架。当用户提供股票代码要求分析投资建议时触发。输入一只股票代码，通过7个SubAgent协作分析（基本面研究员、市场信息研究员、新闻研究员、社交媒体研究员、看涨分析师、看跌分析师、投资组合经理），输出买入/卖出/持有建议及决策依据。触发词："分析这只股票"、"给我投资建议"、"这只票值得... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 227 次。
如何安装 Trading Agents 简化版？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install stock-trading-agents-light」即可一键安装，无需额外配置。
Trading Agents 简化版 是免费的吗？
是的，Trading Agents 简化版 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Trading Agents 简化版 支持哪些平台？
Trading Agents 简化版 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 Trading Agents 简化版？
由 赖根（@laigen）开发并维护，当前版本 v2.4.2。
推荐 Skills