investment-checklist

Codex adapter note

This skill is generated from skills/investment-checklist.md so Claude Code and Codex users share one canonical workflow.

• Treat $ARGUMENTS as the user's request in the current Codex thread.
• When the source mentions Claude-only surfaces such as Task, Agent, WebSearch, Bash, Read, or Write, use the closest Codex capability available in this session: subagents when available, web search when needed, shell commands for local tools, and normal file edits for workspace files.
• Use shared project tools from tools/ in this repository. Prefer running commands from the repository root with paths like python3 tools/financial_rigor.py ...; if the current thread starts outside the repo, locate the actual checkout path first instead of assuming a fixed home-directory path.
• Before starting research, run the date command to confirm today's date; treat it as the baseline for "latest" data and state the data cutoff date in the report header. Never assume the current date from training data.
• Preserve the research quality rules from AGENTS.md: cross-check financial data, use exact arithmetic tools for valuation/math, and clearly label uncertainty and source gaps.

Buffett value investing pre-purchase Checklist

Perform the Buffett value investing pre-purchase Checklist analysis on $ARGUMENTS.

Supported input formats: single or multiple companies, separated by commas/enumeration commas/spaces. For example:腾讯, 茅台, 英伟达 or NVDA AAPL MSFT

Execution flow

Step 1: Parse the input and identify all companies to be analyzed

Parse all company names/tickers from $ARGUMENTS. For each company determine:

• Company full name, stock ticker, listing exchange
• If the company is not listed, mark it as "not listed" and provide a brief explanation (whether there is an indirect investment route), and skip the full Checklist

Step 1.5: AI research bias warning

Perform a quick "information richness" rating (A/B/C) for each company, and note it in the report:

Level	Judgment criteria	Impact on the Checklist
Level A	Listed for many years, abundant data	Execute normally, but beware of the "consensus trap"—all indicators looking clear does not mean it is truly certain
Level B	Limited data requiring estimation	Mark each estimated indicator with a confidence level, and weigh data reliability in the "good business" judgment
Level C	Extremely scarce information	Do not force filling out the six-gate table; honestly mark "insufficient data to judge", and focus on the verifiable core issues

Core principle: The goal of the Checklist is torule out bad choices. For Level C companies, "insufficient data" does not equal "fail", nor does it equal "pass"—it should be honestly marked as "gray area, needs supplementary first-hand information", rather than being judged as rejected just because AI cannot fill out the table.

Duan Yongping said: There are two kinds of "can't understand"—one is that the business is too complex to truly understand, and the other is that you have not yet spent time looking at it. The limitation of AI research is that it easily conflates "little material" with "can't understand".

Step 2: Parallel data collection

Use the Task tool toeach companylaunch an independent background Agent for data collection (all companies launched in parallel at the same time), and each Agent is responsible for collecting:

1. Profitability: ROE (5-10 year trend), gross margin, net margin, free cash flow
2. Valuation data: current stock price, market cap, PE(TTM), forward PE, PB, dividend yield
3. Growth trend: revenue/profit growth rate over the past 3 years
4. Financial health: debt level, capital expenditure needs, cash reserves, net cash/net debt
5. Competitive landscape: market share, main competitors, share change trend
6. Moat evidence: concrete evidence of brand/switching costs/network effects/scale effects/technological barriers
7. Management record: CEO background, key decisions, shareholding, capital allocation record
8. Latest developments: major events in the past 6 months (performance, M&A, regulation, management changes, etc.)

Step 3: Execute the six-gate Checklist company by company

For each listed company, go through the six gates in order:

---

Level 1: Can I understand this business (circle of competence)

Must answer:

• Can I explain in one sentence what this company relies on to make money?
• What business will it most likely still be doing in 10 years?
• Which key variables determine success or failure?
• Does my understanding of this industry come from in-depth research or hearsay?

Scoring criteria（★1-5）：

• ★★★★★: The business model is extremely simple and clear, with high certainty over 10 years (e.g., Moutai: brewing and selling liquor)
• ★★★★☆: The model is clear but has technical barriers and requires some professional knowledge to understand
• ★★★☆☆: The model is understandable but certainty over 10 years is not high, and the industry changes quickly
• ★★☆☆☆: The business lines are complex or the industry is undergoing drastic changes, making the future hard to predict
• ★☆☆☆☆: Completely outside the circle of competence

Hard veto: If you cannot even clearly explain how it makes money, directly mark it as "outside the circle of competence, no analysis".

---

Level 2: Is this a good business (economic characteristics)

Let the data speak,key indicators must be precisely calculated with tools：

python3 tools/financial_rigor.py verify-valuation \
  --price {股价} --eps {EPS} --bvps {每股净资产} --fcf-per-share {每股FCF} --dividend {每股股息}

Indicator	Value for this company	Reference standard	Judgment
ROE (5-year average)		>15% excellent, >20% outstanding
Gross margin		>40% implies pricing power
Free cash flow		Continuously positive, ≈ net profit
Capital expenditure intensity		Asset-light is better than asset-heavy
Debt level		Interest-bearing debt/net profit < 3 years

Scoring criteria（★1-5）：

• ★★★★★: ROE>25%, high gross margin, strong FCF, asset-light, low debt (all meet the standard)
• ★★★★☆: 4 items meet the standard
• ★★★☆☆: 3 items meet the standard
• ★★☆☆☆: 2 items meet the standard or the trend is deteriorating
• ★☆☆☆☆: Most do not meet the standard or FCF is continuously negative

---

Level 3: Is the moat deep enough (competitive advantage)

Check item by item:

Moat type	Whether it exists	Specific evidence	Widening or narrowing?
Brand/pricing power
Switching costs
Network effects
Cost/scale advantage
Technology/patent barriers

Additional test: If you gave a competitor 10 billion, could they replicate this business?

Scoring criteria（★1-5）：

• ★★★★★: Multiple moats叠加 and widening
• ★★★★☆: At least one strong moat and stable
• ★★★☆☆: Has a moat but it is not deep enough, or the trend is unclear
• ★★☆☆☆: The moat is being eroded
• ★☆☆☆☆: No obvious moat

---

Level 4: Is management trustworthy (the human factor)

Check item	Assessment
Honesty (promises vs delivery)
Capital allocation ability (buyback/dividend/M&A record)
Shareholder-interest orientation (shareholding, compensation)
Owner mentality (founder vs professional manager)
Corporate governance (related-party transactions, goodwill, auditing)
Can it still run as usual after the CEO leaves?

Scoring criteria（★1-5）：

• ★★★★★: Founder at the helm, excellent capital allocation, interests fully aligned
• ★★★★☆: Excellent management but with minor flaws
• ★★★☆☆: Management is qualified but there are governance risks
• ★★☆☆☆: Integrity or governance issues
• ★☆☆☆☆: Serious integrity issues (→ hard veto)

---

Fifth gate: Is the price cheap enough (margin of safety)

Metric	Value	Historical percentile	Judgment
PE (TTM)
Forward PE
PB
Dividend yield
FCF Yield

Additional check (Must be calculated precisely with tools; mental math is prohibited）：

python3 tools/financial_rigor.py three-scenario \
  --price {股价} --eps {EPS} --shares {股本亿} \
  --growth {乐观} {中性} {悲观} --pe {乐观PE} {中性PE} {悲观PE} --currency {币种}

• Valuation range under three scenarios (use tool output results)
• If the judgment is wrong, how much could I lose at most by buying at the current price?
• If the stock price halves, would you dare to add to your position?

Scoring criteria（★1-5）：

• ★★★★★: Below 50% of relative intrinsic value, extreme margin of safety
• ★★★★☆: 70% of value, good margin of safety
• ★★★☆☆: Fair valuation, average margin of safety
• ★★☆☆☆: Somewhat expensive, insufficient margin of safety
• ★☆☆☆☆: Severely overvalued

---

Sixth gate: Position sizing and decision discipline (prevent emotional loss of control)

Check the following emotional signals:

• Do you want to buy because of FOMO?
• Do you want to buy only because someone else recommended it?
• If trading were suspended for 5 years, could you accept that?
• Can the buy thesis be written clearly in 200 words or fewer?

---

Step 4: Mirror test

Write a mirror test statement for each company:

"I buy ___ company at ___ yuan, because:

1. The essence of this business is ___, and I understand it;
2. Its moat is ___, and it is widening/narrowing;
3. Management is ___, trustworthy/untrustworthy;
4. The current price is equivalent to ___ discount to intrinsic value, with/without sufficient margin of safety;
5. Even if I am wrong, the downside risk is controllable/uncontrollable, because ___."

If you cannot say it completely in 5 sentences = do not buy. Clearly mark "pass" or "fail".

---

Step 5: Quick veto checklist

Check each company item by item; if any one is triggered, directly mark it as "veto":

• Cannot clearly explain how this company makes money
• Free cash flow has been negative for 3 consecutive years with no visible improvement
• Management has integrity stains
• Competitive advantage is being irreversibly eroded
• Need to rely on "the next greater fool paying a higher price" to make money (greater fool theory)
• Cannot bear the consequences of this investment going to zero
• The main reason for buying is mostly "everyone else is buying" or "it has risen well recently"
• Cannot write the reason for buying clearly in 200 words or fewer

---

Step 6: Output an overview comparison table (required when there are multiple companies)

When analyzing multiple companies, a comparison overview table must be generated:

Company	Checklist passed?	Circle of competence	Good business	Moat	Management	Margin of safety	Core Conclusions
		★☆☆☆☆	★☆☆☆☆	★☆☆☆☆	★☆☆☆☆	★☆☆☆☆

---

Step Seven: Final Conclusions and Write to File

Give a clear conclusion for each company (do not evade):

• ✅ Passed Checklist(X/6 gates) — can proceed to the deep research stage
• ❌ Failed Checklist — explain which red line was triggered
• ❓ Gray area — explain what the key points of dispute are and what investors need to judge for themselves
• N/A — not listed/cannot buy

Write the complete report to ~/巴菲特Checklist-[公司名或"多公司对比"].md

Output format requirements

1. Each company is a separate chapter, including: six-gate scoring table + core data table + key risks (3-5 items) + mirror test + clear conclusion
2. When there are multiple companies, append an overview comparison table at the end
3. All scores must use the ★ symbol (★1-5), without half stars
4. Data must be marked with the source time, and estimated values must be marked as "estimated"
5. End with a concluding remark that echoes Buffett's famous saying: "The first rule of investing is don't lose money"
6. Language style: direct, sharp, no nonsense. Intersperse commentary with quotes from Buffett/Munger/Duan Yongping

Key principles

• Better to miss out than to make a mistake: The goal of the Checklist is to rule out bad choices, not to find the best one
• Be honest about your circle of competence: If you don't understand it, say you don't understand it; don't force an analysis
• Margin of safety is the lifeline: Even a good company bought too expensively will lose money
• The mirror test cannot be skipped: If you can't explain the reason clearly, don't buy, no exceptions