{
  "mcpServers": {
    "boardroom": {
      "command": "npx",
      "args": ["-y", "boardroom-mcp"]
    }
  }
}

"Board: Should I build a mobile app or a progressive web app 
for my new SaaS product?"

.ai/boardroom/
├── LEDGER.md                    # Decision log (institutional memory)
├── BOARD_WISDOM.md              # Distilled principles from past sessions
├── mastermind/
│   ├── SYSTEM_PROMPT.md         # Core operating rules for the boardroom
│   ├── RULE_MATRIX.md           # Which rules fire at each severity
│   ├── COGNITIVE_DOSSIERS.md    # Deep profiles for each advisor
│   ├── SIGNATURE_QUESTIONS.md   # Each advisor's unique probing questions
│   ├── CONVERSATION_TEMPLATE.md # How debate sessions are structured
│   ├── seats/                   # Individual advisor definitions
│   │   ├── 01-ceo.md
│   │   ├── 02-product.md
│   │   ├── 03-marketing.md
│   │   └── ...
│   └── councils/                # Specialized expert panels
│       ├── tech/
│       ├── survival/
│       ├── legal/
│       └── ...

# Seat: Chief Security Officer

## Role
Security & Risk Assessment Lead

## Core Philosophy
"Every system is as secure as its weakest link. 
 Assume breach. Design for resilience."

## Decision Criteria
1. Does this introduce new attack surface?
2. Does this follow the principle of least privilege?
3. Is there a rollback plan if this is compromised?
4. Have we considered supply chain risks?

## Veto Power
Can VETO any action that:
- Exposes PII without encryption
- Removes authentication from an endpoint
- Introduces unaudited third-party dependencies

## Signature Questions
- "What happens if this is compromised at 3 AM?"
- "Who has access, and should they?"
- "What does the blast radius look like?"

## Advice Style
Direct. Data-driven. Uses threat modeling frameworks.
Never says "it depends" without specifying conditions.

.ai/boardroom/mastermind/councils/security/
├── council-config.md        # Council metadata + routing rules
├── seats/
│   ├── cso.md              # Chief Security Officer
│   ├── red-team.md          # Red Team Lead
│   ├── compliance.md        # Compliance Expert
│   └── forensics.md         # Digital Forensics Analyst

# Security & Risk Council

## Routing Keywords
security, authentication, authorization, encryption, 
vulnerability, compliance, GDPR, SOC2, penetration, audit

## Default Severity
standard (elevated to critical for auth/encryption topics)

## Members
- CSO (Chair)
- Red Team Lead
- Compliance Expert
- Digital Forensics Analyst

## Resolution Protocol
Unanimous on security vetoes. Majority on recommendations.

1. Opening Positions
   Each active advisor gives a raw 2-3 sentence gut reaction

2. Interactive Dialogue
   Advisors respond to EACH OTHER:
   **[Member A]**: "Here's why I disagree with B's position..."
   **[Member B]**: "That ignores the data showing..."
   **[Member A] shifts position**: "Actually, if we combine..."

3. Emergent Insights
   Ideas that ONLY emerged from the interaction itself

4. Unresolved Tensions
   Disagreements preserved as data — not forced into consensus

"Board: I want to hear from Young Buffett (1960s, 
aggressive value hunter) AND Late Buffett (2020s, 
quality compounder) on this acquisition."

// The engine invokes two versions of the same advisor
// with different calibration profiles, creating temporal
// debate between past-self and future-self.

"Board: I need expertise on quantum error correction 
for my startup's hardware. None of your councils 
cover quantum computing."

// Prometheus Protocol activates:
// 1. Identifies knowledge gap
// 2. Forges 3-5 temporary advisors calibrated for the domain
// 3. Runs a single-session debate
// 4. Advisor profiles are NOT persisted (fire-and-forget)

┌─────────────────────────────────────────────────┐
│               Your AI Client                     │
│     (Claude, Cursor, Windsurf, OpenClaw, etc.)   │
└─────────────────┬───────────────────────────────┘
                  │ MCP Protocol (STDIO)
                  │ (runs 100% on YOUR machine)
┌─────────────────▼───────────────────────────────┐
│           Boardroom MCP Server                   │
│  ┌──────────┬──────────┬──────────┬───────────┐ │
│  │ analyze  │governance│  intel   │   trust   │ │
│  │          │          │          │           │ │
│  └────┬─────┴────┬─────┴────┬────┴─────┬─────┘ │
│       │          │          │          │        │
│  ┌────▼──────────▼──────────▼──────────▼─────┐ │
│  │           File System Reader               │ │
│  │    (reads .ai/boardroom/ directory)        │ │
│  └────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────┘
                  │
      ┌───────────▼────────────┐
      │  .ai/boardroom/        │
      │  ├── LEDGER.md         │  ← Institutional memory
      │  ├── BOARD_WISDOM.md   │  ← Distilled principles
      │  └── mastermind/       │  ← Advisor definitions
      │      ├── seats/        │
      │      ├── councils/     │
      │      └── protocols/    │
      └────────────────────────┘

# 1. Clone the protocol files
git clone https://github.com/randysalars/boardroom-mcp.git
cp -r boardroom-mcp/protocol-files ~/.ai/boardroom

# 2. Install the MCP server
cd boardroom-mcp && npm install && npm run build

# 3. Register as OpenClaw skill
mkdir -p ~/.openclaw/skills/boardroom
cat > ~/.openclaw/skills/boardroom/SKILL.md << 'EOF'
---
name: boardroom
description: Boardroom Mastermind Council analysis
---
EOF

# 4. Set environment variable
echo 'BOARDROOM_ROOT=~/.ai/boardroom' >> ~/.openclaw/.env

# 5. Register the skill
openclaw skills register boardroom --always