home / skills / openclaw / skills / astock-multiagent-research

astock-multiagent-research skill

/skills/kaneki-jiang/astock-multiagent-research

This skill conducts multi-agent deep research on A-share stocks, delivering multi-perspective analysis, risk checks, and an interactive dashboard for informed

npx playbooks add skill openclaw/skills --skill astock-multiagent-research

Review the files below or copy the command above to add this skill to your agents.

Files (6)
SKILL.md
4.3 KB
---
AIGC:
    ContentProducer: Minimax Agent AI
    ContentPropagator: Minimax Agent AI
    Label: AIGC
    ProduceID: 5f99537637ec8a7284fc6db453befbf1
    PropagateID: 5f99537637ec8a7284fc6db453befbf1
    ReservedCode1: 30450220459a060a3cd402aa1efd0cedaa919091eac38669a1a684f23a3d11e0bbeefab40221009d2ccb1970d6f6ae56965fa89c64a5ce991462be3af54aa3d944f723b80579d9
    ReservedCode2: 3045022072c8a10b604befe2a48fdea3a89439bc7ff3b660ddfa39531dcf9376f3bf2691022100e64cbce383db99fd25a2ea80b7484bbb9e579a3627ad9fc3d5daab163557d93e
description: 多智能体A股深度研究框架。当用户要求分析A股上市公司、产业链研究、个股深度报告、行业比较、投资价值评估时激活。支持:单只股票全面分析(财务/新闻/情绪/技术/风险)、多智能体辩论(看涨vs看跌)、产业链上下游拆解、事实核查、Dashboard部署。数据来源:akshare(A股/港股)、yfinance(美股)、网络搜索(研报/新闻)。适合A股投资者、量化研究员、产业链分析师。
name: astock-multiagent-research
---

# 多智能体A股研究框架

## 快速启动

**单只股票分析**:
```
用户: "帮我研究 600519 茅台" / "分析润泽科技300442" / "中际旭创值得买吗"
```
→ 触发完整6步工作流(见下)

**产业链分析**:
```
用户: "分析AI算力产业链" / "商业航天产业链哪个环节最值得关注"
```
→ 读取 `references/industry-chain.md`,触发产业链专项框架

**快速概览**(仅基本面+新闻):
```
用户: "给我简单看下000977浪潮信息最近怎样"
```
→ 仅启动 fundamentals_analyst + news_analyst 两个子智能体

---

## 工作流(完整6步)

### Step 1:多维并行数据收集

同时启动4个子智能体(使用 `sessions_spawn`):

| 子智能体 | 任务 | 参考提示词 |
|---------|------|-----------|
| `fundamentals_analyst` | 财务报表深度分析、估值、机构预测 | `references/agent-prompts.md#fundamentals` |
| `news_analyst` | 公司动态、行业新闻、政策 | `references/agent-prompts.md#news` |
| `sentiment_analyst` | 市场情绪、研报评级变化 | `references/agent-prompts.md#sentiment` |
| `technical_analyst` | 价格走势、成交量、关键位 | `references/agent-prompts.md#technical` |

### Step 2:观点碰撞

基于Step1报告,并行启动:
- `bullish_researcher`:挖掘增长潜力、竞争优势、价值低估
- `bearish_researcher`:识别风险、业绩隐忧、估值泡沫

### Step 3:风险评估
- `risk_manager`:综合投资风险评估(行业/财务/流动性/政策)

### Step 4:事实核查
- `fact_checker`:对关键数据和说法二次验证

### Step 5:综合报告
按 `references/report-template.md` 格式输出完整研究报告

### Step 6:Dashboard部署
将报告渲染为交互式HTML Dashboard,使用 `deploy` 工具部署,返回可访问链接

---

## 数据获取(Python/akshare)

运行 `scripts/fetch_stock_data.py <股票代码>` 获取基础数据包。

```python
# 手动调用示例
import akshare as ak

# 公司基本信息(A股)
info = ak.stock_individual_info_em(symbol="300442")

# 三大财务报表
balance = ak.stock_balance_sheet_by_report_em(symbol="300442")
income = ak.stock_profit_sheet_by_report_em(symbol="300442")
cashflow = ak.stock_cash_flow_sheet_by_report_em(symbol="300442")

# 财务指标(ROE/毛利率等)
indicator = ak.stock_financial_analysis_indicator(symbol="300442")

# 机构持仓
fund_hold = ak.stock_report_fund_hold(symbol="300442")

# 美股
import yfinance as yf
ticker = yf.Ticker("AAPL")
info = ticker.info
```

子代码格式:沪市6开头用"6xxxxx",深市/创业板用"0xxxxx"/"3xxxxx",北交所用"8xxxxx"。

---

## 输出规范

详见 `references/report-template.md`。核心结构:
- 研究摘要(公司定位、行业地位、当前估值)
- 财务分析表格(营收/净利润/毛利率/ROE,含同比和行业对比)
- 估值分析(PE/PB/PS历史分位)
- 机构观点(评级分布、目标价区间)
- 核心竞争力 vs 风险因素
- 综合评级:🟢高 / 🟡中等 / 🔴低

---

## 注意事项

- akshare股票代码不含市场前缀(用"300442"而非"sz300442")
- 数据来源声明:akshare/东方财富/同花顺/公司公告
- 免责声明必须包含在每份报告末尾
- 产业链分析见 `references/industry-chain.md`

Overview

This skill is a multi-agent research framework for deep A-share (China stock market) analysis. It coordinates specialist agents to produce comprehensive company reports, industry chain breakdowns, multi-perspective debates, risk checks, and an interactive dashboard. It targets A-share investors, quant researchers, and industry analysts who need structured, data-driven insights.

How this skill works

The framework spawns parallel sub-agents to collect financials, news, sentiment, and technical signals, then runs pro/con researcher agents to surface bullish and bearish cases. A risk manager and fact-checker validate findings, and a report assembler formats results into a standardized research template. Finally, the system can render and deploy an interactive HTML dashboard with a persistent link.

When to use it

  • Request a deep-dive single-stock report including financials, news, sentiment, and technical analysis.
  • Analyze an industry chain to identify high-value upstream or downstream nodes.
  • Generate a balanced bull vs. bear debate prior to investment decisions.
  • Produce a validated, shareable research report with risk assessment and disclaimer.
  • Deploy an interactive dashboard for committee review or presentation.

Best practices

  • Provide the stock code (A-share numeric code like 600519 or 300442) and any time horizon or target questions.
  • Specify if you want full workflow (six steps) or a quick overview (fundamentals + news only).
  • Include preferred data sources or date ranges if you need custom backtests or historical context.
  • Treat the output as research support: always run your own compliance and trade checks.
  • Use the dashboard link for collaborative review; keep sensitive credentials out of requests.

Example use cases

  • ’Analyze 600519 Maotai’ — full 6-step report with valuation, risks, and dashboard.
  • ’Map the AI compute industry chain’ — identify key suppliers, customers, and investment nodes.
  • ’Quick look at 000977’ — fundamentals and news summary for an intraday decision.
  • ’Compare two semiconductor names’ — side-by-side financial and valuation comparison.
  • ’Run bull vs. bear debate for a momentum trade’ — highlight catalysts and stop-loss triggers.

FAQ

Which data sources does the skill use?

It relies primarily on akshare for A/H-share data, yfinance for US tickers when needed, and web search for news and research reports.

How do I specify a stock code?

Use the plain numeric A-share code without market prefix (for example, 300442). For US tickers provide the symbol as usual.