eco-moat-ai

name	eco-moat-ai
description	Evaluate any business like Warren Buffett using a 3-stage framework distilled from 49 years of shareholder letters. Supports three modes: quick (gates only), standard (full evaluation), deep (with original letter evidence). Use when: analyzing a company, evaluating a business idea, comparing investment opportunities, screening multiple companies, or when phrases like "is this a good business", "moat", "evaluate", "analyze" appear.
argument-hint	[company name or business description]

eco-moat-ai: Buffett-Grade Business Analysis

Purpose: Evaluate any business through Warren Buffett's lens — 3-stage framework with 28 questions, 7 binary gates, and a 0-100 moat score. Every judgment backed by specific evidence from 49 years of shareholder letters.

Source repo: ~/eco-moat-ai/ (or wherever this repo is cloned)

---

Modes

Detect the mode from the user's prompt, or ask if unclear.

Mode	Trigger Phrases	What You Do	When to Use
Quick	"빠르게", "quick", "screen", "스크리닝", "간단히"	Stage 2 only (7 gates)	Screening 10+ companies fast
Standard	"평가해줘", "evaluate", "analyze" (default)	Full Stage 1-2-3	Evaluating a single company
Deep	"깊이", "deep", "철저히", "원문까지"	Full Stage 1-2-3 + read reference files + cite original letters	Due diligence on a serious candidate

---

Quick Mode

Run only the 7 binary gates. Fast, decisive, no scoring.

1. For each gate, give a one-sentence PASS/FAIL with evidence
2. Output the gate table
3. Verdict: Proceed / Revisit / Reject

# [Company] — Quick Screen

| Gate | Result | Evidence |
|------|--------|----------|
| G1 Understandable | PASS/FAIL | [one sentence] |
| G2 Long-term prospects | PASS/FAIL | [one sentence] |
| G3 Honest management | PASS/FAIL | [one sentence] |
| G4 Sensible price | PASS/FAIL | [one sentence] |
| G5 Demonstrated earnings | PASS/FAIL | [one sentence] |
| G6 Financial independence | PASS/FAIL | [one sentence] |
| G7 No existential leverage | PASS/FAIL | [one sentence] |

**Result**: X/7 -> [Proceed / Revisit / Reject]

---

Standard Mode (Default)

Full 3-stage evaluation with moat scoring.

Stage 1: Intelligence Gathering (28 Questions, 7 Dimensions)

Collect facts before forming any opinion. Answer each question with specific data.

1.1 Industry & Arena

1. Can you explain the business model in one paragraph?
2. Is the industry growing, stable, or declining? 10-year trend?
3. How capital-intensive? Typical capex-to-revenue ratio?
4. How fast does the industry change technologically?

1.2 Competitive Position 5. Market share trend over 5-10 years? 6. Can it raise prices without losing customers? Last time it did? 7. Cost structure vs. competitors? 8. Brand recognition — can consumers identify it unprompted?

1.3 Financial Profile 9. ROE consistently above 15% over 10 years? 10. How much capital needed to maintain current earnings? 11. What does it do with free cash flow? 12. Debt-to-equity? Could it survive 2 years of zero revenue?

1.4 Management Quality 13. CEO skin in the game — net worth in the company? 14. Capital allocation track record? 15. Does management admit mistakes? Search for "mistake"/"error" in reports. 16. Key people retention rate?

1.5 Customer & Product 17. Would customers notice if the company disappeared? 18. How do customers find this company? 19. Customer retention rate? 20. Must-have or nice-to-have?

1.6 Risk Landscape 21. What 3 scenarios would destroy this business? 22. Regulatory risk exposure? 23. Revenue concentration (top customer %, single product %)? 24. Significant leverage?

1.7 Valuation & Context 25. Price-to-owner-earnings vs. 10-year average? 26. What would you pay for the whole business if you could never sell it? 27. Current market sentiment — fearful or greedy toward this sector? 28. Comparable business valuations?

Stage 2: Mental Model Filter (7 Binary Gates — ALL Must Pass)

Gate	Question	Fail = Stop
G1	Can I explain how it makes money in 3 sentences?	"Never invest in a business you cannot understand."
G2	Will it be stronger in 10 years?	"Turnarounds seldom turn."
G3	Do I trust management with my money?	"We've never succeeded in making a good deal with a bad person."
G4	Am I paying below intrinsic value with margin of safety?	"What is smart at one price is stupid at another."
G5	Has it proven earnings through at least one downturn?	"Demonstrated consistent earning power — future projections are of no interest."
G6	Can it survive without external financing?	"Never depend on the kindness of strangers."
G7	Could a bad year kill it?	"It is madness to risk losing what you need in pursuing what you simply desire."

7/7 passed -> Proceed to Stage 3
6/7 passed -> STOP. Can the failing gate be resolved? Revisit in 6 months.
5/7 or below -> REJECT.

Stage 3: Moat Combination Scoring

Score each moat type 0-10:

Moat	What It Means	10 =
M1: Cost Advantage	Structurally lower costs	Widening cost gap competitors cannot close
M2: Brand/Reputation	Customers choose by name, pay premium	Iconic brand with proven pricing power
M3: Switching Cost	Customers can't leave easily	Deep integration, switching is painful
M4: Network Effects	Product improves as more use it	Strong self-reinforcing value loop
M5: Scale Economics	Bigger = permanently cheaper	Dominant, self-reinforcing cost advantage
M6: Regulatory Barrier	Legal/structural protection	Licenses, rights-of-way, natural monopoly

Apply combination multipliers (if both moats score 5+):

Combo	Moats	Multiplier	Why
C1	Cost + Scale	x1.3	Lower costs -> more customers -> even lower costs
C2	Brand + Pricing Power	x1.3	Strong brand funds brand investment, strengthens brand
C3	Switching + Network	x1.4	Can't leave AND product improves. Near-impregnable
C4	Scale + Regulatory	x1.3	Economically impossible for new entrants
C5	Cost + Brand	x1.5	Lowest price AND top-of-mind. Nowhere to attack
C6	Capital-Light + Any Moat	x1.4	Low reinvestment = enormous free cash flow

Final Score: Raw Score x Product of Multipliers x (10/6), capped at 100.

Score	Rating	Action
80-100	Extraordinary	Buy at any fair price. Hold forever.
60-79	Excellent	Buy with margin of safety. Hold for decades.
40-59	Good	Buy only at clear discount. Monitor moat annually.
20-39	Mediocre	Avoid unless extraordinary management + deep discount.
0-19	Avoid	Walk away. No price is low enough.

---

Deep Mode

Everything in Standard Mode, plus:

1. Read reference files before evaluating. Map each dimension to its deep-dive:

Checklist Section	Read This File
1.1 Industry & Arena	business-reference/03-cost-advantage-and-scale.md
1.2 Competitive Position	business-reference/01-competitive-moats.md, business-reference/02-pricing-power-and-brand.md
1.3 Financial Profile	business-reference/04-capital-allocation.md
1.4 Management Quality	business-reference/05-management-quality.md
1.5 Customer & Product	business-reference/02-pricing-power-and-brand.md
1.6 Risk Landscape	business-reference/06-risk-leverage-survival.md
1.7 Valuation & Context	business-reference/07-acquisition-discipline.md

2. Cite original letters by reading markdown/buffett-letter-YYYY.md when a specific principle needs Buffett's exact words
3. Compare to Buffett's actual cases — reference GEICO, See's, or failed acquisitions from case-studies/ where parallels exist
4. Quote Buffett directly at least 3 times in the report, with year and context

Deep Mode output adds these sections to the Standard report:

## Buffett Parallels
[Which Berkshire company or case study is most similar? Why?]

## Key Letters Referenced
| Year | Relevant Principle | How It Applies |
|------|-------------------|----------------|
| [YYYY](markdown/buffett-letter-YYYY.md) | [principle] | [application to this company] |

---

Output Format (Standard & Deep)

# [Company Name] — Buffett Business Evaluation

**Mode**: Standard / Deep
**Date**: [evaluation date]

## Stage 1: Intelligence Summary
[Key facts organized by the 7 dimensions. Be specific — numbers, trends, sources.]

## Stage 2: Gate Results
| Gate | Result | Evidence |
|------|--------|----------|
| G1 Understandable | PASS/FAIL | [one sentence] |
| G2 Long-term prospects | PASS/FAIL | [one sentence] |
| G3 Honest management | PASS/FAIL | [one sentence] |
| G4 Sensible price | PASS/FAIL | [one sentence] |
| G5 Demonstrated earnings | PASS/FAIL | [one sentence] |
| G6 Financial independence | PASS/FAIL | [one sentence] |
| G7 No existential leverage | PASS/FAIL | [one sentence] |

**Result**: [X/7 gates passed -> Proceed / Stop / Reject]

## Stage 3: Moat Score
| Moat Type | Score (0-10) | Evidence |
|-----------|-------------|----------|
| M1 Cost Advantage | X | [why] |
| M2 Brand/Reputation | X | [why] |
| M3 Switching Cost | X | [why] |
| M4 Network Effects | X | [why] |
| M5 Scale Economics | X | [why] |
| M6 Regulatory Barrier | X | [why] |

Raw Score: X/60
Applicable Combos: [list with multipliers]
**Final Moat Score: X/100 — [Rating]**

## Verdict
[2-3 sentence Buffett-style summary. Reference specific letter principles where applicable.]

---

Batch Screening Mode

When the user provides a list of companies:

1. Run Quick Mode for all companies first
2. For those that pass 7/7, run Standard Mode
3. Output a comparison table:

| Company | G1-G7 | Moat Score | Rating | Key Moat | Key Risk |
|---------|-------|-----------|--------|----------|----------|
| Co A | 7/7 | 82 | Extraordinary | Cost+Scale | Regulatory |
| Co B | 7/7 | 45 | Good | Brand | Tech disruption |
| Co C | 5/7 | -- | Rejected | -- | Failed G3, G7 |

4. Rank by moat score and provide commentary on top candidates.

---

File Map

All paths are relative to the repo root (~/eco-moat-ai/).

Purpose	Path
Evaluation framework	buffett-business-checklist.md
Practical how-to guide	buffett-field-guide.md
Deep-dive evidence (7 files)	business-reference/*.md
Worked case studies	case-studies/*.md
Original letters (49 files)	markdown/buffett-letter-YYYY.md
Year-by-year lessons (49 files)	lessons/lessons-YYYY.md
Theme analyses (16 files)	themes/*.md
Wisdom synthesis	buffett-business-wisdom-synthesis.md
Quote collection	buffett-quotations-collection.md
Korean checklist	ko/buffett-business-checklist.md
Korean field guide	ko/buffett-field-guide.md

---

Installation

Clone the repo and symlink into your Claude skills:

git clone https://github.com/juliuschun/eco-moat-ai.git ~/eco-moat-ai
ln -s ~/eco-moat-ai/.claude/skills/eco-moat-ai ~/.claude/skills/eco-moat-ai

Or copy just the skill folder if you prefer.