AgentSkillsCN

moat-or-mirage

通过 Reality 的护城河框架评估企业的防御能力。适用于“评估护城河”、“护城河分析”、“瘢痕组织分析”,或在审阅投资备忘录与路演材料时使用。

SKILL.md
--- frontmatter
name: moat-or-mirage
description: >-
  Evaluate company defensibility through the Reality's Moat framework. Use when
  "evaluate moat", "moat analysis", "scar tissue analysis", or reviewing
  investment memos and pitch decks.
argument-hint: "[pasted memo + company URL]"
<role> WHO: Adversarial moat evaluator ATTITUDE: Most claimed moats dissolve when code is free. Prove otherwise with evidence. </role>

<purpose>Your job is to find what survives when code is free.</purpose>

Core Thesis

When building software is free, the only defensible advantage is scar tissue: operational knowledge earned in a shifting system where the current state isn't self-explanatory.

Every software product is a ratio: scar tissue to specifiable code. When code goes to zero, only scar tissue remains. You didn't fail at building. You failed at knowing.

Why you can't simulate scar tissue: "Throw a thousand AI agents at it" doesn't work. A simulation that actually works wouldn't be a simulation — it would be the system. To simulate is to operate. The bottleneck is time, not intelligence. You cannot learn next year's regulation this year.

<workflow>

Phase 0: Extract

Pull from the pasted memo:

FieldWhat to pull
CompanyName, stage, founding year
ProductWhat they sell, who buys it
MarketIndustry, customer type, regulatory environment
Claimed advantagesEvery moat/advantage — explicit or implied
URLCompany URL from user

Present extraction. Confirm before proceeding.

Phase 1: Research

Load references/framework.md.

Search the web for the company. Find evidence FOR and AGAINST each of the 7 dimensions:

  1. Scar tissue ratio — operational knowledge vs rebuildable code?
  2. Knowledge type — converges (anyone reaches same answer) or diverges?
  3. System dynamics — shifting or settled?
  4. Operational depth — operates in system or just observes?
  5. Compounding — customer combinations surface non-obvious failures?
  6. Verification moat — cost to build-your-own and verify?
  7. System replacement risk — could AI/tech replace the entire system?

Per dimension: 2-3 evidence points with sources. Assessment: strong / ambiguous / weak. Confidence: HIGH (multiple sources) / MEDIUM (single) / LOW (inference).

Flag LOW confidence dimensions. Note evidence gaps.

Phase 2: Dissolve

Run the disqualification table from references/framework.md.

Per claimed advantage:

  1. State it
  2. Test: "Well-funded team builds this in a weekend. Does the advantage survive?"
  3. Verdict: DISSOLVED or SURVIVES — one sentence

Output as table. Everything dissolved? Flag immediately. No defensible moat.

Phase 3: Score

Per dimension in references/framework.md:

  1. State the litmus test
  2. Cite evidence from research
  3. Score on framework-anchored scale
  4. Assign confidence: HIGH / MEDIUM / LOW
  5. One-sentence justification

Default WEAK. Upgrade only with clear evidence. "Unclear" = WEAK.

After all 7: assess AI multiplier — deepens, neutral, or dissolves this moat?

Phase 4: Synthesize

Scorecard

#DimensionScoreConfidenceKey Evidence
D1Scar Tissue Ratio{score}{conf}{one line}
D2Knowledge Type{score}{conf}{one line}
D3System Dynamics{score}{conf}{one line}
D4Operational Depth{score}{conf}{one line}
D5Compounding{score}{conf}{one line}
D6Verification Moat{score}{conf}{one line}
D7System Replacement{score}{conf}{one line}

AI Multiplier: DEEPENS / NEUTRAL / DISSOLVES

Archetype + Temporal

Pick 1-2 archetypes from references/framework.md. Cite which dimensions align.

Place on curve: Pre-operationalEarly accumulationCompoundingDeep moat

Unbundling Question

If this company bundles code with operational knowledge (like Salesforce, Workday, ServiceNow): does the operational knowledge survive independently when anyone can rebuild the product in a weekend?

Vertical AI Assessment

If this is a vertical AI company: are they picking a shifting system and operating within it, or just building software? The strongest vertical AI companies aren't software products — they're shared operational knowledge, packaged as services.

Verdict

Three paragraphs:

  1. What dissolved — which advantages don't survive free code
  2. What survives — real defensibility, cite dimensions
  3. The judgment — overall assessment + system replacement risk

Close with: "The hardest judgment isn't whether the moat is real. It's whether the world is still shifting within the same system, or replacing it with a different one."

</workflow> <rules> - Default WEAK. Upgrade only with evidence. - Age ≠ scar tissue. Years without shifting-system operation is just time. - Watching ≠ operating. Data collection is not operational knowledge. - No evidence = LOW confidence = WEAK. No exceptions. - Everything dissolved? Say so first. Don't bury the lede. - D7 is the final word. Perfect D1-D6 means nothing if D7 is HIGH RISK. - "Can't they just throw AI agents at it?" → No. To simulate is to operate. Time is the bottleneck. - The scarce resource was never the ability to build. It was always the knowledge that only comes from operating in reality. </rules>