stock-research-executor

股票投资调研执行引擎，执行8阶段投资尽调流程。接收stock-question-refiner生成的结构化调研指令，部署多智能体并行研究，生成带引用的投资尽调报告。覆盖：公司事实底座、行业周期、业务拆解、财务质量、股权治理、市场分歧、估值护城河、综合报告。当用户需要进行股票投资研究、基本面分析、投资尽调时使用此技能。

349stars80forksUpdated 7/16/2026

#stock-analysis #fundamental-analysis #finance #due-diligence #multi-agent #investment-research

Security Assessment

Safe(100/100)

Security Score100/100

About stock-research-executor

Stock-research-executor is an investment due-diligence execution engine that turns structured research prompts into comprehensive, well-cited fundamental analysis reports. It solves the problem of conducting rigorous, repeatable equity research by running a standardized 8-phase framework covering business foundation, industry cycle analysis, business breakdown, financial quality, ownership and governance, market sentiment (bull and bear cases), valuation and moat, and a final synthesis report. It is designed to receive a structured investment research prompt from the companion stock-question-refiner skill and transform it into an objective, fact-based due diligence report rather than investment advice.

The skill deploys a multi-agent research strategy, launching multiple Task agents in parallel for each phase to improve efficiency, then coordinates and synthesizes their findings while preserving source attribution. It uses tools including Task, WebSearch, WebFetch, the mcp__web_reader__webReader extractor, Read, Write, and TodoWrite. Each phase produces a structured markdown report with an executive summary, detailed findings, key data tables, a source-quality assessment (A-E grading), and noted contradictions or gaps. The workflow verifies the structured prompt (ticker, market, investment style, holding period, scope), creates an execution plan, deploys parallel agents for phases 1-7, performs quality assurance such as citation verification and profit-versus-cash-flow cross-validation, and generates a final synthesis with a signal-light rating and monitoring checklist. Outputs follow a standardized directory structure under RESEARCH/STOCK_[ticker]_[company]/.

The skill targets investors and analysts performing fundamental analysis and investment due diligence across A-share, Hong Kong, and US markets, with support for value, growth, turnaround, and dividend styles. It emphasizes objectivity, requiring every factual claim to carry a citation (author, date, title, URL) and distinguishing facts from opinion. A stated requirement is a complete structured prompt from stock-question-refiner; if inputs are incomplete the skill asks for clarification before proceeding, and it explicitly does not provide investment advice.

FAQ

What input does this skill need to start?

It expects a complete structured research prompt (typically from the stock-question-refiner skill) containing the stock ticker and company name, market (A-share/HK/US), investment style, holding period, research scope across the 8 phases, output requirements, and research constraints. If the prompt is incomplete, the skill asks for clarification before proceeding.

What are the 8 phases of the research process?

Phase 1 Business Foundation, Phase 2 Industry Analysis, Phase 3 Business Breakdown, Phase 4 Financial Quality, Phase 5 Governance Analysis, Phase 6 Market Sentiment (bull and bear cases), Phase 7 Valuation and Moat, and Phase 8 Final Synthesis producing the investment research report.

How does it use multiple agents?

For each phase it deploys multiple Task agents in parallel (a single message with multiple tool calls), assigning each agent a specific research focus and then compiling, de-duplicating, and synthesizing their findings while preserving source attribution. The skill explicitly instructs never to launch agents sequentially.

Does the skill give buy or sell recommendations?

No. It is built to maintain objectivity and produce facts over narratives, explicitly avoiding investment advice. The final report uses a signal-light rating and a monitoring checklist rather than a recommendation.

How are sources and citations handled?

Every factual claim must carry a citation in the format author, date, title, URL, and each source is graded on an A-E quality scale. Quality assurance includes citation verification, profit-versus-cash-flow cross-validation, and company-versus-peer comparison before final synthesis.