メニュー
ACTIVATE for ANY finance, investment, trading, or market query. Comprehensive value investing framework combining Buffett, Munger, Duan Yongping, and Li Lu methodologies. Use when making investment decisions.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/oyi77/1ai-skills --skill wolf-financeコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Android and mobile application security testing — emulators, rooting, traffic interception, dynamic instrumentation. Use when testing mobile apps for vulnerabilities, reversing APKs, or bypassing security controls on Android.
Self-reflection + Self-criticism + Auto-learning from corrections + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when working with self improving.
Plan and execute a comprehensive red team engagement covering reconnaissance through post-exploitation using MITRE ATT&CK-aligned TTPs to evaluate an organization's detection and response capabilities. Use when working with conducting full scope red team engagement.
SOC 職業分類に基づく
SKILL.md を表示中
| name | wolf-finance |
| description | ACTIVATE for ANY finance, investment, trading, or market query. Comprehensive value investing framework combining Buffett, Munger, Duan Yongping, and Li Lu methodologies. Use when making investment decisions. |
| domain | financial |
| license | Apache-2.0 |
| tags | ["analysis","crypto","finance","investment","testing","trading","wolf","money","value-investing"] |
| version | 2.0.0 |
| author | oyi77 |
| subdomain | |
| type | finance |
Systematic value investing compounds at 15-25% annually. A $50K portfolio following this framework generates $7.5K-12.5K/year. With position sizing and margin of safety, drawdowns stay under 20%. One good investment thesis can return 2-10x over 3-5 years.
#!/usr/bin/env bash
# Portfolio health check
mkdir -p ~/wolf-finance/{holdings,research,theses,reviews}
echo "=== Portfolio Health Check ==="
echo "1. List ALL current holdings with cost basis"
echo "2. Run 7-factor Quality Screen (eliminate weak positions):"
echo " - Declining revenue (3yr) -> RED FLAG"
echo " - Increasing debt/equity -> RED FLAG"
echo " - Negative free cash flow -> RED FLAG"
echo " - Insider selling >50% -> RED FLAG"
echo " - Losing market share -> RED FLAG"
echo " - Regulatory risk -> RED FLAG"
echo " - Overvalued (P/E > 3x industry) -> RED FLAG"
echo "3. Flag any position with 3+ red flags for review"
echo "4. Check thesis: has anything fundamentally changed?"
echo "5. Decision: HOLD, INCREASE, or EXIT each position"
| Excuse | Truth |
|---|---|
| "I need more information before deciding" | You have enough to decide. More info = more noise. |
| "The market is too volatile right now" | Volatility is when value investors buy |
| "I should wait for a better entry" | DCA in. Time in market > timing the market |
On completion: "Portfolio: [N] positions, $[N] value, [N] red flags, [N] thesis updates needed, [N] actions taken"
Trigger phrases:
"wolf finance"
"Use when working with wolf finance"
Analyzing any financial asset (equities, crypto, forex, commodities, derivatives)
Building investment theses with evidence-tiered backing (T1/T2/T3)
Running pre-trade risk gates before position entry
Portfolio construction and risk management across asset classes
Institutional-grade reporting for investment committees
Deep company research and valuation
Management quality assessment and due diligence
Quality screening of investment candidates
Wolf Finance provides finance operations with accuracy and compliance. Integrates comprehensive investment research frameworks drawn from Buffett, Munger, Duan Yongping, and Li Lu — from quality screening through deep company research to final decision memorandum.
# Example: Portfolio risk calculation
def calculate_risk(returns: list[float]) -> dict:
import statistics
mean = statistics.mean(returns)
stdev = statistics.stdev(returns)
sharpe = mean / stdev if stdev > 0 else 0
return {"mean": mean, "stdev": stdev, "sharpe_ratio": sharpe}
| Rationalization | Reality |
|---|---|
| "The market will recover" | Do not hope. Analyze. Set stop-losses and follow your strategy. |
| "I do not need to track expenses" | What you do not measure, you cannot optimize. Track everything. |
| "One spreadsheet is enough" | Financial models need version control and audit trails. Use proper tools. |
Every key financial data point MUST come from two independent sources. Flag discrepancies > 1%.
Data source priority by market:
| Market | Primary Source | Secondary Source | Raw Filings |
|---|---|---|---|
| US stocks | macrotrends.net | stockanalysis.com | SEC EDGAR 10-K/Q |
| HK stocks | aastocks.com | macrotrends (ADR) | HKEX disclosure |
| A-share | eastmoney.com | cninfo.com.cn | CNINFO original PDF |
Error calculation:
Error rate = |source1 - source2| / source1 x 100%
| Error | Action |
|---|---|
| <= 1% | Use source1 value, cite both |
| 1-5% | Flag discrepancy, note possible cause (exchange rate, accounting method) |
| > 5% | Must check original filing — do not use |
Common discrepancy causes: GAAP vs Non-GAAP, exchange rate timing, fiscal year definitions, consolidation scope, data lag.
Mandatory calculation verification — use tooling, never LLM mental math:
python3 tools/financial_rigor.py verify-market-cap --price {price} --shares {shares} --reported {reported_mcap} --currency {currency}python3 tools/financial_rigor.py verify-valuation --price {price} --eps {eps} --bvps {bvps} --fcf-per-share {fcf} --dividend {dividend}python3 tools/financial_rigor.py cross-validate --field {field} --values '{"source1": val, "source2": val}' --unit {unit}python3 tools/financial_rigor.py three-scenario --price {price} --eps {eps} --shares {shares} --growth {opt} {base} {pess} --pe {opt_pe} {base_pe} {pess_pe} --years 3An 8-module systematic investment research process combining the frameworks of Buffett, Munger, Duan Yongping, and Li Lu.
Before starting, rate the company's "AI-researchability" to identify potential data bias:
| Grade | Feature | AI Research Trap | Strategy |
|---|---|---|---|
| A (info-rich) | Listed long time, heavy analyst coverage | Consensus too strong — AI output converges to market pricing, limited alpha | Focus on devil's advocate: why don't smart people buy? What risks are overlooked? |
| B (moderate info) | Listed 1-3yr, limited coverage | AI fills gaps with "reasonable estimates" — looks complete but gives false certainty | Label confidence on every estimated data point |
| C (info-scarce) | New listing/obscure/emerging market, near-zero coverage | AI over-cautious due to lack of data — misreads as "unclear = bad" | First-principles mode: focus on core business questions, not report completeness |
Bias self-check (maintain throughout research):
Collect using parallel research agents:
Cross-validate every key data point using the Financial Data Standards section above and the financial_rigor.py tool suite. Never rely on LLM mental arithmetic for calculations.
Common error prevention:
Evaluate each moat type:
| Moat Type | Verification Method |
|---|---|
| Brand / pricing power | Can it raise price without losing volume? |
| Switching costs | How expensive is it for customers to switch? |
| Network effects | Does the product get better with more users? |
| Scale advantages | How significant are cost advantages from scale? |
| Technology / patent barriers | How many years ahead? Can it be replicated? |
Analyze moat trend: wider or narrower in the past 5 years? Outlook for the next 5 years.
Buffett-style question: Will this moat still exist in 10 years? What could destroy it?
Summary table:
| Dimension | Conclusion | Confidence |
|---|---|---|
| Business quality (Duan Yongping) | ||
| Moat (Buffett) | ||
| Management (Duan + Buffett) | ||
| Biggest risk (Munger) | ||
| Civilization trend (Li Lu) | ||
| Valuation (Buffett + Duan) |
Decision table:
| Strategy | Recommendation |
|---|---|
| Not holding | |
| Holding | |
| Sell signal | |
| Add signal |
Report must include:
tools/report_audit.pyA 6-gate checklist adapted from the Buffett-Munger framework. Each gate scored *1-5.
Hard veto: if you cannot clearly describe how it makes money, mark "outside competence — no analysis."
| Metric | Company Value | Reference Standard | Pass? |
|---|---|---|---|
| ROE (5yr avg) | > 15% good, > 20% excellent | ||
| Gross margin | > 40% suggests pricing power | ||
| Free cash flow | Consistently positive, near net income | ||
| Capex intensity | Asset-light > asset-heavy | ||
| Debt level | Interest-bearing debt / net profit < 3yr |
Scoring: *1-5 based on how many standards are met and whether trends are improving or deteriorating.
Check each type:
| Moat Type | Evidence | Widening or Narrowing? |
|---|---|---|
| Brand / pricing power | ||
| Switching costs | ||
| Network effects | ||
| Cost / scale advantage | ||
| Technology / patent barriers |
Additional test: If a competitor had $10B, could they replicate this business?
| Check | Assessment |
|---|---|
| Honesty (promises vs delivery) | |
| Capital allocation (buyback/dividend/M&A track record) | |
| Shareholder alignment (ownership, compensation) | |
| Owner mindset (founder vs professional manager) | |
| Governance (related-party transactions, goodwill, audit) | |
| Would the company run well without the CEO? |
Hard veto: severe integrity issues.
| Metric | Value | Historical Percentile | Assessment |
|---|---|---|---|
| PE (TTM) | |||
| Forward PE | |||
| PB | |||
| Dividend yield | |||
| FCF Yield |
Three-scenario valuation — MUST use tooling, never mental math:
python3 tools/financial_rigor.py three-scenario \
--price {price} --eps {eps} --shares {shares} \
--growth {opt} {base} {pess} --pe {opt_pe} {base_pe} {pess_pe} --currency {currency}
Write:
"I am buying ___ at ___ because:
- The business essence is ___, I understand it;
- Its moat is ___ and it is widening/narrowing;
- Management is ___ worthy/unworthy of trust;
- Current price represents ___ of intrinsic value, ___ sufficient/insufficient margin of safety;
- Even if I am wrong, downside is controllable/uncontrollable because ___."
Cannot complete in 5 sentences = do not buy.
Quickly filter out non-first-class companies using 7 hard criteria with 3 exemption rules.
| # | Metric | Elimination Condition | What It Measures |
|---|---|---|---|
| 1 | 10yr avg ROE | < 8% | Capital efficiency — can equity beat opportunity cost? |
| 2 | 5yr cumulative FCF | Negative | Real cash generation vs paper profits |
| 3 | Interest coverage (EBIT/interest) | < 2x | Debt repayment safety |
| 4 | Long-term gross margin | < 15% | Pricing power — product differentiation |
| 5 | Operating CF / Net income (5yr avg) | < 0.7 | Profit quality — can profit be collected as cash? |
| 6 | Long-term net margin | < 5% | Resilience — does revenue volatility zero out profit? |
| 7 | 5yr share dilution | > 20% (non-M&A) | Shareholder interest — is management diluting you? |
A: Strategic Investment Period (exempts Rule 1)
B: Active Low-Margin Strategy (exempts Rule 6)
C: High-Turnover Thin-Margin Model (exempts Rules 4 and 6)
Passing screen does not equal "good investment" — it means the company survived first-pass elimination. Further research on business model sustainability, management, valuation, and competitive dynamics is still required.
Deep management quality assessment when standard management scoring is uncertain (*** or below) or management is the core investment thesis.
Identify CEO, CFO, founder (if not CEO), controlling shareholder, other key executives. Distinguish "who makes decisions" from "who has the title."
Strategic vision: Review CEO public statements (letters to shareholders, earnings calls, interviews, social media) over the past 5 years. Extract predictions and compare with actual outcomes.
| Time | CEO's Judgment/Prediction | Actual Outcome | Accuracy |
|---|
Execution ability:
| Dimension | Assessment | Evidence |
|---|---|---|
| Strategy to execution | Did they deliver what they said? | |
| Organization | Can they attract and retain talent? | |
| Crisis handling | How did they respond to difficulties? | |
| Iteration speed | How fast do they correct mistakes? |
Promise vs delivery tracking — extract specific commitments from past 3 years:
| # | Time | Commitment | Venue | Delivered? | Score |
|---|
Track record:
| Fulfillment Rate | Rating |
|---|---|
| > 80% | Excellent — say what they do |
| 60-80% | Acceptable — right direction, execution gap |
| 40-60% | Concerning — over-promise, under-deliver |
| < 40% | Severe issue — cannot be trusted |
Crisis behavior: Search for the company's major crises. How did management react?
Stakeholder attitudes:
| Stakeholder | Attitude | Evidence |
|---|---|---|
| Shareholders | Respect / ignore / exploit | |
| Employees | Good treatment / exploitation / indifference | |
| Customers | Customer-centric / short-term extraction | |
| Regulators | Compliance / gray-area play |
M&A record:
| Time | Target | Amount | Strategic Logic | Result | Score(1-5) |
|---|
Buyback record: Check valuation at time of buyback vs current. Dividend record: Payout ratio vs FCF, sustainability. New business investment: Area, cumulative spend, current status, return.
Scoring:
| Dimension | Score(1-5) | Notes |
|---|---|---|
| M&A discipline | Right price? Integration success? | |
| Buyback timing | Buying low, stopping high? | |
| Dividend rationality | Payout matches FCF? | |
| New business investment | Success rate, stop-loss discipline | |
| Cash management | Reasonable reserves vs hoarding? |
| Question | Answer |
|---|---|
| Can the company run normally if CEO leaves tomorrow? | |
| Management team depth — clear successor? | |
| Competitive advantage dependent on CEO or organization/systems? | |
| Historical CEO transitions — smooth? |
| Dimension | Weight | Score(1-5) | Weighted |
|---|---|---|---|
| Integrity | 35% | ||
| Strategy & Execution | 25% | ||
| Capital allocation | 25% | ||
| Governance | 15% | ||
| Composite | 100% |
Portfolio-level investment thesis tracking, drift analysis, and structured portfolio review. These systems are modeled after the disciplined buy-and-hold processes used by Buffett, Li Lu, and Duan Yongping — methods that transform portfolio management from reactive guessing into systematic oversight.
The thesis tracker is a buy-and-hold discipline system that enforces documented reasoning before entry and systematic re-validation through quarterly check-ins.
Design principle: Most investors stop at research -> buy -> pray. Missing post-entry tracking causes reluctance to sell, panic-selling on drawdowns, and forgetting why you bought. The system answers one question at every check: Would you still buy this today if you did not own it?
Two modes:
Mode A — Build Thesis:
tools/financial_rigor.py verify-valuationreports/{company}-thesis.mdMode B — Track Thesis:
10 - (black_countx3) - (red_countx2) - (yellow_countx1) - (redline_triggersx5), min 1 max 10Key principles for thesis tracking:
Separates genuine evidence changes from price noise and stylistic rewording by comparing two thesis snapshots across five fixed dimensions.
Design principle: The hardest thing about holding long-term positions is distinguishing:
Three modes:
Mode A — Compare specified reports:
tools/financial_rigor.py — no LLM mental mathMode B — Auto snapshot comparison:
Find old and new thesis snapshots in reports/{company}-thesis*.md; verify same company, different dates; execute Mode A.
Mode C — Missing baseline handling: State explicitly that drift detection cannot run without a historical baseline. Guide user to first build a thesis via the Thesis Tracker. Do not fabricate an old thesis from memory or market impression.
Key principles for drift detection:
A structured 7-step portfolio review process that treats portfolio management as a separate discipline from stock-picking.
Design principle: Researching companies is only half of investing. The other half is portfolio-level decisions: position sizing, funding source (new money vs swap), correlation with existing holdings, opportunity cost — every dollar should go where it earns the most.
Seven steps:
Step 1: Parse positions — Normalize input holdings into standard table (ticker, quantity, cost basis, current price, market value, weight, P&L). Support both proportional and unit formats. Load saved portfolio if available (reports/portfolio-latest.md).
Step 2: Get current data — Use parallel sub-agents via WebSearch for each position: current price and valuation (PE, PB, dividend yield), latest quarter financial changes, major recent events, analyst consensus. Validate with tools/financial_rigor.py verify-valuation.
Step 3: Single-position health check — For each position answer three questions:
Step 4: Portfolio-level analysis:
Step 5: Optimization suggestions:
Step 6: Output report with executive summary covering: portfolio health rating (Outstanding / Good / Needs Attention / Severe), single most important action to take, biggest current risk
Step 7: Save portfolio file to reports/portfolio-latest.md with holdings table, review date, rebalance log, next review reminder
Key principles for portfolio review: