菜单
Natural language queries to the knowledge graph. Search past reports, screenings, trades, research, and market context.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Investment note management. Record, retrieve, and delete investment theses, concerns, and lessons.
Deep research on stocks, industries, markets, and business models. Integrates Grok API (X/Web search) and yfinance for multi-angle analysis reports.
Plan Mode — Orchestrator oversees workflow design, execution, autonomous loop, and review. Activated when told "in plan mode."
Undervalued stock screening. EquityQuery-based screening without a predefined stock list. Searches for undervalued stocks across 60+ regions including Japan, US, ASEAN, Hong Kong, Korea, and Taiwan using P/E, P/B, dividend yield, ROE, and more.
Portfolio management. Display holdings, record trades, and analyze portfolio structure. Input data foundation for stress tests.
Detailed report for individual stocks and ETFs. Generates a financial analysis report from a ticker symbol. Individual stocks show valuation, undervaluation score, and shareholder return rate. ETFs show expense ratio, AUM, and fund size.
基于 SOC 职业分类
| name | graph-query |
| description | Natural language queries to the knowledge graph. Search past reports, screenings, trades, research, and market context. |
| argument-hint | Natural language query (e.g., What was the last report on Toyota?) |
| allowed-tools | Bash(python3 *) |
Search past data accumulated in the knowledge graph (Neo4j) using natural language.
python3 .claude/skills/graph-query/scripts/run_query.py "natural language query"
For natural language → skill selection, see .claude/rules/intent-routing.md.
Results are displayed in Markdown format. If no data is found, a message is shown. If Neo4j is not connected, displays "No data found."
When displaying graph query results, integrate with conversation context: