adversarial-analysis

Adversarial Analysis

RULE: Before implementing any significant proposal, systematically attack it to find weaknesses.

The Principle

Proposals that survive adversarial scrutiny are more robust. This skill provides a 6-phase framework for stress-testing ideas before commitment.

Loading This Skill

Agents can load via:

required_skills:
  - adversarial-analysis

Or dynamically invoke:

skill: adversarial-analysis

---

The 6-Phase Framework

Phase 1: Pre-mortem

"It's 6 months from now. This failed. Why?"

Assume complete failure and work backwards. What went wrong?

Process:

1. Vividly imagine the failure state
2. Identify 5-7 distinct failure scenarios
3. Trace each back to root causes
4. Note which causes were preventable

Output format:

### Pre-mortem
> "It's 6 months from now. This failed. Why?"
- [Failure scenario 1]: [Root cause]
- [Failure scenario 2]: [Root cause]
- [Failure scenario 3]: [Root cause]

---

Phase 2: Assumption Audit

Every proposal rests on assumptions. Surface and stress-test them.

Process:

1. List every assumption (explicit and implicit)
2. Rate confidence: H (high) / M (medium) / L (low)
3. For each, answer: "If wrong, what breaks?"

Output format:

### Assumptions (confidence: H/M/L)
| Assumption | Confidence | If Wrong |
|------------|------------|----------|
| [Assumption 1] | M | [Impact if false] |
| [Assumption 2] | L | [Impact if false] |
| [Assumption 3] | H | [Impact if false] |

Prioritize low-confidence, high-impact assumptions for deeper analysis.

---

Phase 3: Edge Case Storm

Generate 10+ scenarios that could break the proposal.

Process:

1. Think about boundary conditions
2. Consider rare but possible inputs
3. Imagine hostile actors
4. Consider scale extremes (0, 1, many, millions)
5. Consider timing/ordering issues
6. Consider resource exhaustion
7. Consider integration failures

Output format:

### Edge Cases
- [Breaking scenario 1]
- [Breaking scenario 2]
- [Breaking scenario 3]
... (minimum 10)