新上了个买方研报 skill，直接从 sec 官方拉原始文件分析，写得真的太好了，现在仓位变动都要先过一下 skill 心里才有底。 买方研报 skill 已开源，地址： https://github.com/haskaomni/serenity-skill… 研报地址： https://app.k2ai.dev/s/US5T0zOVJp_obEOJcYvqZXG0qtmVHYoJkynB-3Q…

haskaomni/serenity-skill

Source: https://github.com/haskaomni/serenity-skill

Serenity Skills

Codex skills for translating market information into testable investment research frameworks.

Skills

• serenity-alpha: translates market news into alpha hypotheses using a news -> demand -> financial statements -> small-cap elasticity -> validation path framework.
• bayesian-intrinsic-growth-valuation: estimates a company’s intrinsic 3-5 year growth rate with Bayesian hypothesis updates, then compares it with market-implied growth and FOMO.
• gf-dma-health-index: scores whether a stock’s current valuation/trend health is supported by fundamental growth speed, DMA trend speed, divergence, escape ratio, and estimate revisions.
• tam-adj-peg: evaluates growth-stock valuation by adjusting traditional PEG with TAM runway and business quality.
• buy-side-equity-research-memo: generates source-backed buy-side equity research memos from a ticker, with investment view, SEC/IR-backed financial analysis, valuation scenarios, catalysts, risks, and Serenity framework cross-checks.

直接使用托管版

如果你觉得本地安装、配置 Codex skill 或维护环境不方便，也可以订阅 @iamai_omni，然后访问 app.k2ai.dev 直接使用托管版。订阅版不需要你自己搭建，并且会附赠许多其他功能，适合想快速上手、持续使用 Serenity 体系的用户。也可以扫码直接打开订阅页：

Repository Layout

skills/
├── serenity-alpha/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/original-framework.md
├── bayesian-intrinsic-growth-valuation/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/original-framework.md
├── gf-dma-health-index/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/original-framework.md
├── tam-adj-peg/
│   ├── SKILL.md
│   ├── agents/openai.yaml
│   └── references/original-framework.md
└── buy-side-equity-research-memo/
    ├── SKILL.md
    ├── agents/openai.yaml
    └── references/original-framework.md

Each subdirectory under skills/ is an independent Codex skill. Codex discovers a skill from its SKILL.md; files under references/ are supporting material loaded only when needed.

Install

Copy all skills into your Codex skills folder:

mkdir -p "${CODEX_HOME:-$HOME/.codex}/skills"
cp -R skills/* "${CODEX_HOME:-$HOME/.codex}/skills/"

Or install only one skill:

mkdir -p "${CODEX_HOME:-$HOME/.codex}/skills"
cp -R skills/serenity-alpha "${CODEX_HOME:-$HOME/.codex}/skills/"
cp -R skills/bayesian-intrinsic-growth-valuation "${CODEX_HOME:-$HOME/.codex}/skills/"
cp -R skills/gf-dma-health-index "${CODEX_HOME:-$HOME/.codex}/skills/"
cp -R skills/tam-adj-peg "${CODEX_HOME:-$HOME/.codex}/skills/"
cp -R skills/buy-side-equity-research-memo "${CODEX_HOME:-$HOME/.codex}/skills/"

Then invoke $serenity-alpha for news-to-alpha analysis, $bayesian-intrinsic-growth-valuation for Bayesian intrinsic-growth valuation, $gf-dma-health-index for trend/valuation health scoring, $tam-adj-peg for TAM-adjusted PEG valuation, or $buy-side-equity-research-memo for a full buy-side stock memo. If a newly copied skill does not appear, restart Codex.

What They Do

serenity-alpha:

• Separates narrative news from already-observable demand changes.
• Maps demand into revenue, margin, cash-flow, and balance-sheet impact.
• Searches for small, pure, potentially misclassified beneficiaries.
• Builds 1-4 quarter verification chains and falsification points.
• Frames position posture conditionally as research, not personalized investment advice.

bayesian-intrinsic-growth-valuation:

• Converts fundamentals, industry cycle, TAM, valuation, and new information into H0-H5 growth-hypothesis probabilities.
• Updates 3-5 year revenue CAGR assumptions with Bayesian reasoning instead of surface bullish/bearish labels.
• Separates intrinsic growth updates from FOMO, narrative heat, and valuation multiple expansion.
• Compares weighted intrinsic growth with market-implied growth.
• Classifies valuation as undervalued, fair, expensive but tradable, or bubble-like.

gf-dma-health-index:

• Combines revenue growth, profit growth, estimate revisions, and 20/50/100/200DMA structure.
• Scores fundamental-DMA match, price-DMA divergence, trend parallelism, and revision confirmation.
• Classifies the current state from healthy momentum to broken/escaping.

tam-adj-peg:

• Adjusts traditional PEG with TAM Runway Factor and Quality Factor.
• Separates growth speed from growth duration, TAM capture, pricing power, cyclicality, dilution, and execution risk.
• Classifies valuation from very cheap to very expensive and maps it to core, high-beta, turnaround, option-like, or cyclical position framing.

buy-side-equity-research-memo:

• Starts with rating bias, target-price range, upside/downside, key debate, and thesis breakpoint.
• Uses SEC filings, company IR, earnings calls, presentations, and other current sources to anchor key facts.
• Builds industry-chain, competitive-position, financial-statement, value-driver, SOTP/valuation, and Bull/Base/Bear scenario sections.
• Integrates Serenity Alpha, Bayesian Intrinsic Growth, TAM-Adj-PEG, and GF-DMA lenses only when they improve the investment decision.
• Lists catalysts, risks, variant perception, monitoring dashboard, and source list for follow-up research.

License

MIT