red-blue-validator

What this skill does

# Red/Blue Team Validator

> "Find weaknesses before reality does."

Every proposition—whether a decision, strategy, architecture, or plan—has vulnerabilities. This skill systematically exposes them through iterative adversarial cycles. Red Team attacks with substantive, steel-manned challenges. Blue Team defends with mitigations and hardening. The cycle continues until convergence: a battle-tested proposition with documented defenses.

---

## 1. Purpose

### Core Value Proposition

Static analysis misses what adversarial pressure reveals. Red/Blue validation simulates the attacks your proposition will face—from competitors, critics, reality itself—and forces you to build defenses before you need them. The output is not just a risk list, but a hardened proposition that has survived systematic assault.

### Capabilities

| # | Capability | Phase | Value |
|---|-----------|-------|-------|
| 1 | Proposition intake with attack surface mapping | Pre-Round | Define what can be attacked |
| 2 | Experience pool loading (domain failure patterns) | Pre-Round | Avoid reinventing known failures |
| 3 | Multi-category attack generation | Round N: Red | Surface vulnerabilities systematically |
| 4 | Steel-manning attacks to maximum strength | Round N: Red | Ensure attacks are not strawmen |
| 5 | Severity scoring (CRITICAL/HIGH/MEDIUM/LOW) | Round N: Red | Prioritize responses |
| 6 | Defense generation (REFUTE/MITIGATE/ACCEPT/HARDEN) | Round N: Blue | Address each attack |
| 7 | Proposition hardening through iterative refinement | Round N: Blue | Strengthen against attacks |
| 8 | Convergence evaluation with explicit criteria | Round N: Eval | Know when to stop |
| 9 | RISK-ASSESSMENT synthesis (CONTRACT-08) | Post-Round | Standardized output |
| 10 | Hardened proposition generation | Post-Round | Battle-tested version |
| 11 | Attack/defense log compilation | Post-Round | Audit trail |
| 12 | Go/no-go recommendation | Post-Round | Decision support |

---

## 2. When to Use

### Ideal Use Cases

| Scenario | Why Red/Blue Validation Matters |
|----------|--------------------------------|
| **Pre-commitment decision review** | Simulate objections before committing resources |
| **Strategy validation** | War-game competitive responses and market realities |
| **Architecture decision hardening** | Stress-test technical choices before implementation |
| **Proposal defense preparation** | Anticipate and prepare for stakeholder pushback |
| **Investment due diligence** | Adversarial review of financial projections and market assumptions |
| **Security posture assessment** | Systematic attack surface enumeration |
| **Go/no-go decisions** | High-stakes decisions need adversarial pressure |
| **Policy/process validation** | Find edge cases and failure modes |
| **Product launch readiness** | Anticipate market, competitive, and operational challenges |
| **M&A target evaluation** | Adversarial review of synergy claims |

### Anti-Patterns (When NOT to Use)

| Anti-Pattern | Why It's Ineffective | Better Alternative |
|--------------|----------------------|-------------------|
| **Low-stakes decisions** | Over-engineering for trivial choices | Just decide and iterate |
| **Time-critical emergencies** | Fires need extinguishing, not philosophy | Act, then debrief |
| **Already committed** | Adversarial review after commitment creates conflict | Use for future decisions |
| **Early exploration** | Premature to attack ideas still forming | Use after initial validation |
| **Confirmation theater** | Going through motions without genuine adversarial intent | Either commit to true adversarial thinking or skip |
| **Reversible decisions** | Two-way doors don't need siege testing | Save intensity for one-way doors |

---

## 3. Parameters

| Parameter | Type | Required | Default | Description |
|-----------|------|----------|---------|-------------|
| `subject_type` | enum | **yes** | — | decision \| strategy \| architecture \| plan \| policy \| investment \| security |
| `max_rounds` | integer | no | 3 | Maximum red/blue cycles (1-5) |
| `attack_intensity` | enum | no | standard | light \| standard \| aggressive |
| `attack_categories` | list | no | auto | Categories to probe (see catalog); auto selects by subject_type |
| `convergence_mode` | enum | no | no_new_critical | no_new_critical \| all_addressed \| round_limit |
| `include_experience_pool` | boolean | no | true | Load domain-specific failure patterns |
| `steel_manning_level` | enum | no | standard | minimal \| standard \| maximum |
| `output_mode` | enum | no | full | risk_assessment \| hardened_proposition \| full_log |

### Parameter Effects Matrix

| Parameter | Red Phase Effect | Blue Phase Effect | Convergence Effect |
|-----------|-----------------|------------------|-------------------|
| `attack_intensity: light` | Top 3 attack categories | Quick defenses | max_rounds capped at 2 |
| `attack_intensity: standard` | Top 5 attack categories | Full defense protocol | Normal convergence |
| `attack_intensity: aggressive` | All applicable categories | Exhaustive defense | Requires no_new_critical |
| `steel_manning_level: minimal` | 1-pass attacks | — | Faster rounds |
| `steel_manning_level: standard` | 2-pass steel-manning | — | Normal rounds |
| `steel_manning_level: maximum` | 3-pass with ideological Turing test | — | Thorough rounds |
| `convergence_mode: no_new_critical` | — | — | Stop when 0 new CRITICAL/HIGH |
| `convergence_mode: all_addressed` | — | Must address all | Stop when no ACCEPT responses |
| `convergence_mode: round_limit` | — | — | Stop at max_rounds |

### Auto-Selected Attack Categories by Subject Type

| Subject Type | Default Attack Categories |
|--------------|--------------------------|
| `decision` | ASSUMPTIONS, ALTERNATIVES, REVERSIBILITY, CONSEQUENCES, TIMING |
| `strategy` | COMPETITIVE, MARKET, EXECUTION, DEPENDENCIES, TIMELINE |
| `architecture` | SCALABILITY, SECURITY, DEPENDENCIES, OPERATIONAL, EDGE_CASES |
| `plan` | FEASIBILITY, RESOURCES, TIMELINE, DEPENDENCIES, RISKS |
| `policy` | EDGE_CASES, ENFORCEMENT, UNINTENDED_CONSEQUENCES, POLITICAL |
| `investment` | ECONOMIC, MARKET, EXECUTION, COMPETITIVE, ASSUMPTIONS |
| `security` | ATTACK_SURFACE, VULNERABILITIES, DEPENDENCIES, OPERATIONAL |

---

## 4. Checkpoints

This skill uses interactive checkpoints (see `references/checkpoints.yaml`) to resolve ambiguity:
- **subject_type_classification** — When proposition type is ambiguous
- **attack_intensity_selection** — When attack intensity not specified
- **convergence_mode_selection** — When convergence criteria not specified
- **premature_convergence_check** — When convergence met but warning signs present
- **infinite_loop_risk** — When defenses generate more attacks than they resolve
- **output_mode_selection** — When output format not specified

---

## 5. Iterative Workflow

### Workflow Overview

```
┌─────────────────────────────────────────────────────────────────────────────┐
│                       RED/BLUE TEAM VALIDATOR                               │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                              │
│  ╔══════════════════════════════════════════════════════════════════════╗   │
│  ║                      PRE-ROUND SETUP                                  ║   │
│  ║  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐                   ║   │
│  ║  │ Proposition │  │   Attack    │  │ Experience  │                   ║   │
│  ║  │   Intake    │─▶│  Surface    │─▶│    Pool     │                   ║   │
│  ║  │             │  │  Mapping    │  │   Loading   │                   ║   │
│  ║  └─────────────┘  └─────────────┘  └─────────────┘                   ║   │
│  ╚══════════════════════════════════════════════════════════════════════╝   │
│                                  │                                           │
│                                  ▼