stock-analysis
This skill should be used when the user asks to analyze this stock, find a target price or buy zone, review fundamentals or technicals, assess risk/earnings/dividends, screen stocks, analyze sector rotation, build an ETF portfolio, design an options strategy, or analyze the macro market; also for Chinese requests such as "分析这只股票", "给出买入区间或目标价", "做基本面/技术面/风险/财报/股息分析", "筛选股票", "分析板块轮动", "构建 ETF 组合", "设计期权策略", "分析宏观市场"; or when they provide a stock ticker, security, ETF, portfolio, earnings event, or investment thesis that requires source-backed market research and an actionable decision.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/arthur19940725--stock-analysis-skill)HTML
<a href="/listings/arthur19940725--stock-analysis-skill"><img src="/api/listings/arthur19940725--stock-analysis-skill/badge.svg" alt="findsafeskills safety rating" /></a>Install
Install instructions from the README
As written by the repo's author. findsafeskills has not verified these steps.
Copy this repository into the Agent's skills directory, for example:
~/.agents/skills/stock-analysis/
Then describe the task in an Agent that supports skills:
Use stock-analysis to analyze Apple Inc. (Nasdaq: AAPL) over a 12-month horizon. Write the report in English. Use stock-analysis to review the fundamentals and valuation of NVIDIA Corp. (Nasdaq: NVDA). Write the report in English. Use stock-analysis to assess the technical setup, support levels, and risks of SPDR S&P 500 ETF Trust (NYSE Arca: SPY). Write the report in English. Use stock-analysis to design a maximum-loss-defined AAPL options strategy using a synchronized bid/ask chain. Write the report in English.
The prompts above demonstrate invocation syntax. They are not frozen datasets or reproducible investment cases.
The skill can also be invoked explicitly:
$stock-analysis Analyze Microsoft Corp. (Nasdaq: MSFT) over a 12-month horizon. Write the report in English.
Before a full analysis, provide the ticker, primary exchange, holding period, benchmark, risk tolerance, and current position status when possible. Missing fields are handled under [`references/research-protocol.md`](references/research-protocol.md) and must be disclosed.
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 129 lines
- ---
- name: stock-analysis
- description: This skill should be used when the user asks to analyze this stock, find a target price or buy zone, review fundamentals or technicals, assess risk/earnings/dividends, screen stocks, analyze sector rotation, build an ETF portfolio, design an options strategy, or analyze the macro market; also for Chinese requests such as "分析这只股票", "给出买入区间或目标价", "做基本面/技术面/风险/财报/股息分析", "筛选股票", "分析板块轮动", "构建 ETF 组合", "设计期权策略", "分析宏观市场"; or when they provide a stock ticker, security, ETF, portfolio, earnings event, or investment thesis that requires source-backed market research and an actionable decision.
- ---
- # Stock Analysis
- 把股票、ETF、组合、板块、期权与宏观请求转换为可核验、可复算、可证伪的研究结论。围绕一个决策问题组织完整单股研究:
- > 在指定持有期内,以当前可验证价格持有该证券,其情景加权总回报是否超过要求回报,并足以补偿永久损失、事件风险和证据不确定性?
- ## 先选择运行模式
- | 用户意图 | 运行模式 | 加载文件 |
- |---|---|---|
- | “全面分析”“是否值得买”“目标价”“完整投资报告” | 单股总控 | `references/research-protocol.md`、`references/decision-model.md`、`frameworks/00-thesis-driven-research.md`、`references/output-quality.md` |
- | “基本面”“估值”“商业模式”“护城河” | 专项研究 | `references/research-protocol.md` + `frameworks/01-coverage-memo.md` |
- | “技术面”“买入区间”“支撑/阻力”“RSI/MACD” | 专项研究 | `references/research-protocol.md` + `frameworks/02-technical-panel.md` |
- | “风险”“回撤”“压力测试”“对冲” | 专项研究 | `references/research-protocol.md` + `frameworks/03-risk-framework.md` |
- | “财报”“业绩”“EPS”“guidance” | 专项研究 | `references/research-protocol.md` + `frameworks/04-earnings-analyzer.md` |
- | “股息”“分红”“收益率”“DRIP” | 专项研究 | `references/research-protocol.md` + `frameworks/05-dividend-analyzer.md` |
- | “板块”“轮动”“行业配置” | 专项研究 | `references/research-protocol.md` + `frameworks/06-sector-rotation.md` |
- | “筛选”“量化”“多因子”“Top 10” | 专项研究 | `references/research-protocol.md` + `frameworks/07-quant-screener.md` |
- | “ETF 组合”“资产配置”“再平衡”“定投” | 专项研究 | `references/research-protocol.md` + `frameworks/08-etf-portfolio.md` |
- | “期权”“价差”“covered call”“iron condor” | 专项研究 | `references/research-protocol.md` + `frameworks/09-options-architect.md` |
- | “宏观”“利率”“通胀”“美联储”“市场状态” | 专项研究 | `references/research-protocol.md` + `frameworks/10-macro-outlook.md` |
- | 财报/公告后更新、持仓复核 | 增量更新 | 原研究结论 + `frameworks/00-thesis-driven-research.md` 的增量流程 |
- 命中多个专项时,只加载真正改变决策的文件。完整单股研究不得把多个 framework 并行生成后直接拼接;严格按总控依赖链执行。
- ## 输入核验
- 先核验证券身份、主交易所、证券类型、报价币种、分析截止时间和持有期。遇到会实质改变结论的歧义时,只问一个最关键问题并停止;否则显式列出默认值后继续。
- 至少取得或可靠核验:
- - `ticker` 与 `exchange`;
- - 当前价格 `P0`、币种、价格时间戳和来源;
- - 最近正式财务披露及发布日期;
- - 用户任务类型和期望持有期。
- 完整单股评级还需要基准、最低要求回报、三情景价值桥、风险压力情景和证据评分。输入字段与默认值见 `references/research-protocol.md`。
- ## 取证与数据纪律
- 1. 优先读取用户提供的文件、模型与链接;把用户假设与外部事实分开。
- 2. 优先监管申报、交易所公告、公司 IR、正式财报、官方宏观数据和产品发行人页面。
- 3. 再使用有方法说明的可靠数据源;财经媒体和聚合行情只作补充或价格来源。
- 4. 把搜索摘要、论坛和社交媒体仅作为寻找原始资料的线索。
- 5. 给每个会改变结论的市场数字附 `as of`;不得把错时价格、共识、利率、IV 或财务数据伪装成同步快照。
- 6. 先解析精确标的再调用行情工具。验证市场、证券类型、合约/现货属性与币种,避免同名 ticker 错配。
- 7. 若当前环境没有某个特定 MCP,直接使用可用的行情连接器、交易所/发行人页面或可靠网页;不要因固定工具名不存在而停止。
- 8. 在可访问来源全部失败后才向用户索取当前价。不得凭记忆估算当前价、估值倍数或期权报价。
- 对关键事实使用 `[F]`,外部共识 `[C]`,分析估计 `[E]`,假设 `[A]`,计算推导 `[D]`,无法核验 `[U]`。使用 `K01…` 标识可证伪主张,使用 `S01…` 标识来源。
- ## 数据闸门与降级
- 缺失数据时降低输出精度,不补造数字:
- - 无完整价格序列:不给 RSI、MACD、波动率、beta 或最大回撤,只列所需 OHLCV、参数和观察计划。
- - 无点时共识:不用今天的网页数字倒填历史预期,改用公司指引与透明自建估计。
- - 无同步双边期权链:只给参数化结构和取链清单,不给具体合约、权利金、Greeks、胜率或“可执行”标签。
- - 无历史点时股票池/退市样本/真实发布日期:不声称完成回测,不报 Alpha、Sharpe 或胜率。
- - 无资金流数据:不把成交量、持仓或 ETF 价格变化称为机构净流入。
- - 无可靠情景概率或价值桥:输出条件估值或暂不评级,不套用固定 20/60/20。
- ## 单股总控执行
- 加载 `frameworks/00-thesis-driven-research.md`,按以下有向依赖执行:
- 1. `M10 → M4`:确定宏观状态、基准、行业利润池与竞争约束。
- 2. `M4 → M1 → M8 → M3`:建立经营现实、现金回报与验证日历。
- 3. `M2-A → M5`:形成条件估值/反向估值草案,再用因子、盈利修正和价格隐含预期挑战。
- 4. `M7 → M2-B`:把风险回写到经营假设、估值参数、情景概率和压力情景,再形成最终目标价。
- 5. `M6 + M9 → 总控`:制定现货执行;仅在期权闸门通过时设计期权实施;最后统一计算评级。
- 只允许总控层给最终评级。各模块仅输出证据、反证、模型输入和 HANDOFF。若 M6/M9 揭示重大新风险,最多回查一次 `M7 → M2-B`;仍未解决则降低证据分或暂不评级。
- ## 确定性计算
- 把三情景、要求回报、证据分和否决项写入 JSON,调用:
- ```bash
- python scripts/decision_math.py path/to/input.json
- ```
- 正斜杠路径在 Windows 与 Unix 的 Python 上均可使用;Windows 也可用 `path\to\input.json`。
- 使用 `scripts/decision_math.py` 计算 `TR_s`、`ETR`、`EU`、`ED`、`PCR`、`SL`、`H`、`ECS` 与机械评级。禁止手工改变阈值以配合叙事。输入结构、公式和评级门槛见 `references/decision-model.md`。
- 专项任务若没有完整三情景和证据基础,只给该专项能支持的“技术倾向、风险状态、条件估值、候选清单或参数化结构”,不要冒充完整股票评级。
- ## 输出
- 默认用简体中文,保留 ticker、指标、机构名和行业术语的英文。短问题先给行动结论,再给依据和失效条件;完整研究使用 `references/output-quality.md` 模板。
- 所有行动建议至少包含:
- - 数据截止时间与来源层级;
- - 当前行动或观察状态;
- - 触发/确认条件;
- - 技术止损与基本面失效分别列示;
- - 目标、风险回报与持有期;
- - 数据缺口及其对结论的影响。
- 结尾添加:
- > 本报告仅供一般信息与研究用途,不构成个性化投资、法律或税务建议。
- ## 资源索引
- - `references/research-protocol.md`:统一输入、来源、同步、标签、降级与研究预算。
- - `references/decision-model.md`:情景回报、证据评分、机械评级和脚本输入。
- - `references/output-quality.md`:完整报告模板、检查清单、自动失败与更新纪律。
- - `frameworks/00-thesis-driven-research.md`:十模块总控依赖、HANDOFF 和专项工作流。
- - 专项框架(与上方路由表一致):
- - `frameworks/01-coverage-memo.md`
- - `frameworks/02-technical-panel.md`
- - `frameworks/03-risk-framework.md`
- - `frameworks/04-earnings-analyzer.md`
- - `frameworks/05-dividend-analyzer.md`
- - `frameworks/06-sector-rotation.md`
- - `frameworks/07-quant-screener.md`
- - `frameworks/08-etf-portfolio.md`
- - `frameworks/09-options-architect.md`
- - `frameworks/10-macro-outlook.md`
- - `scripts/decision_math.py`:确定性评级计算与输入验证。
- - `scripts/test_decision_math.py`:核心边界与降级测试。
README.md · README · 177 lines
- # Stock Analysis Skill
- English | [中文速览](#中文速览)
- ## Overview
- A source-first securities research skill for AI agents. It supports stocks, ETFs, sector rotation, earnings, dividends, technical analysis, risk, options, and macro research.
- Its purpose is not to produce plausible-sounding investment conclusions. It organizes research into a verifiable, reproducible, and falsifiable decision process.
- This README is English-first. The skill currently generates Simplified Chinese by default; add `Write the report in English` to a request when English output is required.
- ## Quick start
- Copy this repository into the Agent's skills directory, for example:
- ```text
- ~/.agents/skills/stock-analysis/
- ```
- Then describe the task in an Agent that supports skills:
- ```text
- Use stock-analysis to analyze Apple Inc. (Nasdaq: AAPL) over a 12-month horizon. Write the report in English.
- Use stock-analysis to review the fundamentals and valuation of NVIDIA Corp. (Nasdaq: NVDA). Write the report in English.
- Use stock-analysis to assess the technical setup, support levels, and risks of SPDR S&P 500 ETF Trust (NYSE Arca: SPY). Write the report in English.
- Use stock-analysis to design a maximum-loss-defined AAPL options strategy using a synchronized bid/ask chain. Write the report in English.
- ```
- The prompts above demonstrate invocation syntax. They are not frozen datasets or reproducible investment cases.
- The skill can also be invoked explicitly:
- ```text
- $stock-analysis Analyze Microsoft Corp. (Nasdaq: MSFT) over a 12-month horizon. Write the report in English.
- ```
- Before a full analysis, provide the ticker, primary exchange, holding period, benchmark, risk tolerance, and current position status when possible. Missing fields are handled under [`references/research-protocol.md`](references/research-protocol.md) and must be disclosed.
- ## How it works
- The skill first verifies the security identity, price timestamp, currency, and data sources. It then selects the relevant framework. A full single-stock review follows this dependency chain:
- ```text
- Macro and industry context
- -> Operations, financials, and catalysts
- -> Conditional valuation and market checks
- -> Risk audit and stress testing
- -> Execution plan and final rating
- ```
- Core controls:
- - **Source-first**: prefer regulatory filings, exchange notices, company investor relations, and official data.
- - **Evidence labels**: separate verified facts `[F]`, consensus `[C]`, estimates `[E]`, assumptions `[A]`, derivations `[D]`, and unknowns `[U]`.
- - **Data gates**: reduce precision when data is missing instead of inventing prices, indicators, consensus estimates, or option quotes.
- - **Scenario valuation**: calculate expected total return, stress loss, and payoff coverage across bear, base, and bull cases.
- - **Deterministic rating**: use [`scripts/decision_math.py`](scripts/decision_math.py) and fixed thresholds instead of changing a rating to fit a narrative.
- - **Falsifiable thesis**: attach supporting evidence, the strongest counterevidence, and an invalidation condition to each core claim.
- ## Reproducible stock-analysis practice
- This practice uses a public issuer, a fixed filing, and explicit verification steps. It is a frozen fundamental-data exercise, not a current recommendation.
- ### Research snapshot
- | Field | Fixed value |
- |---|---|
- | Security | Apple Inc. common stock |
- | Identifiers | Nasdaq: `AAPL`; SEC CIK `0000320193` |
- | Research cutoff | 2025-10-31 16:00 ET (`America/New_York`) |
- | Financial period | FY2025, 2024-09-29 through 2025-09-27 |
- | Filing | Form 10-K filed 2025-10-31; accession `0000320193-25-000079` |
- | Financial units | USD millions, except per-share data |
- | Market-price rule | Retrieve the 2025-10-31 regular-session close separately and record its exact field definition |
- ### Public sources
- | Source ID | Publisher and source | Publication / access date | What to verify |
- |---|---|---|---|
- | S01 | SEC, [filing detail for accession 0000320193-25-000079](https://www.sec.gov/Archives/edgar/data/320193/000032019325000079/0000320193-25-000079-index.htm) | Published 2025-10-31; accessed 2026-07-31 | Issuer identity, form, filing date, report period, accession, and filing documents |
- | S02 | SEC, [Apple FY2025 Form 10-K](https://www.sec.gov/Archives/edgar/data/320193/000032019325000079/aapl-20250927.htm) | Published 2025-10-31; accessed 2026-07-31 | Audited financial statements, product and Services sales, EPS, and cash-flow line items |
- | S03 | SEC, [Company Facts JSON](https://data.sec.gov/api/xbrl/companyfacts/CIK0000320193.json) and [EDGAR API documentation](https://www.sec.gov/search-filings/edgar-application-programming-interfaces) | Live API; accessed 2026-07-31 | Programmatic cross-check of standard US-GAAP facts; this is another representation of S02, not independent corroboration |
- | S04 | Apple Investor Relations, [historical stock-price lookup](https://investor.apple.com/stock-price/default.aspx) and [FAQ](https://investor.apple.com/investor-relations/faq/default.aspx) | Dynamic; accessed 2026-07-31 | Historical price lookup; Apple states that the quote feed is supplied by Ticker Technologies and delayed by 20 minutes |
- | S05 | Nasdaq, [AAPL historical quotes](https://www.nasdaq.com/market-activity/stocks/aapl/historical) | Dynamic; accessed 2026-07-31 | Historical OHLCV and the regular-session price field when the public feed is available |
- ### Verified facts and calculations
- The following filing values were checked against S02. Standard entity-level facts were also reconciled to S03.
- | Metric | FY2025 | FY2024 | Evidence |
- |---|---:|---:|---|
- | Total net sales | 416,161 | 391,035 | `[F]` S02/S03 |
- | Services net sales | 109,158 | 96,169 | `[F]` S02 |
- | Net income | 112,010 | 93,736 | `[F]` S02/S03 |
- | Diluted EPS (USD/share) | 7.46 | 6.08 | `[F]` S02/S03 |
- | Cash generated by operating activities | 111,482 | 118,254 | `[F]` S02/S03 |
- | Payments for acquisition of property, plant, and equipment | 12,715 | 9,447 | `[F]` S02/S03 |
- Reproducible derivations:
- - Total net-sales growth = `416,161 / 391,035 - 1` = **6.43%** `[D]`.
- - Services net-sales growth = `109,158 / 96,169 - 1` = **13.51%** `[D]`.
- - FY2025 Services mix = `109,158 / 416,161` = **26.23%** `[D]`.
- - FY2025 net margin = `112,010 / 416,161` = **26.92%** `[D]`.
- - FY2025 free-cash-flow proxy = `111,482 - 12,715` = **USD 98,767 million** `[D]`.
- The free-cash-flow proxy is the skill's default calculation of operating cash flow less purchases of property, plant, and equipment. It is not an Apple-reported GAAP line item.
- ### How to verify the practice
- 1. Open S01 and match `Apple Inc.`, CIK `0000320193`, Form 10-K, report date `2025-09-27`, filing date `2025-10-31`, and accession `0000320193-25-000079`.
- 2. In S02, find `Products and Services Performance`, `Consolidated Statements of Operations`, and `Consolidated Statements of Cash Flows`; reconcile the table above before calculating any ratio.
- 3. In S03, filter standard XBRL facts by `form=10-K`, `fy=2025`, `fp=FY`, `end=2025-09-27`, and `accn=0000320193-25-000079`. Useful concepts include:
- - `us-gaap:RevenueFromContractWithCustomerExcludingAssessedTax`
- - `us-gaap:NetIncomeLoss`
- - `us-gaap:EarningsPerShareDiluted`
- - `us-gaap:NetCashProvidedByUsedInOperatingActivities`
- - `us-gaap:PaymentsToAcquirePropertyPlantAndEquipment`
- 4. Do not take the last JSON array item without filtering: later filings and comparative periods can repeat the same fact. Verify Services sales in the S02 table because dimensional product data may not appear as an entity-level Company Facts value.
- 5. For the historical market price, query 2025-10-31 in S04 and S05. Record:
- ```text
- ticker | exchange | price field | value | currency | trading date |
- timestamp and timezone | delayed status | adjustment definition |
- source URL | retrieval time
- ```
- 6. Use the regular-session close for historical `P0`; do not silently substitute an adjusted close or an intraday/after-hours quote. If only one feed works, disclose that the price lacks a public cross-check. If neither feed works or a discrepancy cannot be resolved, label `P0` as `[U]` and do not issue a price target or final stock rating.
- 7. Keep all later information out of the frozen 2025-10-31 analysis to avoid look-ahead bias.
- A reproducible prompt:
- ```text
- Use stock-analysis to review Apple Inc. common stock (Nasdaq: AAPL;
- SEC CIK 0000320193) using only information public by 2025-10-31
- 16:00 ET. Reconcile the FY2025 facts and calculations in the README
- to sources S01-S03. Retrieve the 2025-10-31 regular-session price
- from S04 and S05 when available and record its currency, timestamp,
- delay, and adjustment definition. Disclose a missing cross-check; if
- neither source works or they conflict, mark P0 as [U] and stop before
- a target price or final rating. Write the report in English.
- ```
- The verified snapshot supports a limited operating observation: Services grew faster than total sales in FY2025, while the stated free-cash-flow proxy was USD 98.767 billion. It does not, by itself, supply forecast scenarios, required return, benchmark-relative return, or a defensible buy/hold/sell rating.
- ## Repository layout
- ```text
- stock-analysis/
- ├── SKILL.md # Skill entry point and routing
- ├── frameworks/ # Research controller and 10 specialist frameworks
- ├── references/ # Data protocol, rating model, and output specification
- └── scripts/
- ├── decision_math.py # Deterministic rating calculation
- └── test_decision_math.py # Boundary and degradation tests
- ```
- Run the tests:
- ```bash
- python scripts/test_decision_math.py
- ```
- ## Disclaimer
- This project is for general information and research only. It is not personalized investment, legal, or tax advice. Market data is time-sensitive; recheck every conclusion against the latest primary sources before acting.
- ## 中文速览
- 这是一个面向 AI Agent、以证据和数据闸门为核心的证券研究 skill,覆盖股票、ETF、行业轮动、财报、股息、技术面、风险、期权与宏观市场。README 以英文为主;skill 当前默认输出简体中文,如需英文报告,请在任务中明确写入 `Write the report in English`。
- 安装时将仓库复制到 Agent 的 skills 目录,然后提供 ticker、主交易所、持有期、基准、风险承受能力和当前持仓状态。完整规则见 [`SKILL.md`](SKILL.md) 与 [`references/research-protocol.md`](references/research-protocol.md)。
- 上面的实践使用 Apple Inc.(Nasdaq: AAPL;CIK `0000320193`)FY2025 Form 10-K 的固定公开数据,并给出 SEC 原文、Company Facts、Apple IR 与 Nasdaq 行情的具体核验路径。财务事实与动态行情分开处理;若历史价格无法公开核验,必须把 `P0` 标为 `[U]`,停止输出目标价和最终评级。
- 本项目仅供一般信息与研究用途,不构成个性化投资、法律或税务建议。
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.