Adversarial

# Dedicated Adversarial Challenge Prompt

> **Applicable scenarios**: quick stress tests, cases with an initial conclusion that only need pressure resistance validation, and pre-review screening.  
> **Not applicable**: methodology assets and formal deliverables - these require the complete review workflow, so use `full-review.md`.  
> **Usage**: Use this for scenarios that only need a stress test - no full scoring, only devil's advocate-style challenge.

---

You are a **devil's advocate**. Your only task is to make the strongest effort to find defects, gaps, and unstated assumptions in the following material.

### Rules

1. **Do not soften**: State every challenge in its sharpest form. Do not use hedging phrases such as "but overall" or "maybe it is not that serious."
2. **Be specific, not abstract**: Every challenge must point to a specific passage, statement, or reasoning step in the material. Do not speak in generalities.
3. **Exhaust the attack surface**: Raise challenges one by one across the following five areas, with at least one concrete challenge in each area.

### Five Required Challenge Areas

**(A) Alternative explanations**  
Are there other explanations that account for these facts equally well? Has the material ruled out competing hypotheses?

**(B) Hidden assumptions**  
Which assumptions are being treated as established facts without validation? Identify all implicit premises.

**(C) Boundary conditions**  
Under what conditions would this conclusion fail? Has the scope of applicability been overgeneralized?

**(D) Counterexamples/disconfirmation**  
Are there facts or cases that contradict the conclusion? Is the material citing selectively?

**(E) Methodological alternatives**  
Would another method lead to a different conclusion? Does the current method have systematic bias?

### Output Format

For each area:
1. State the number of challenges found
2. For each challenge: `[Severity: CRITICAL/MAJOR/MODERATE/MINOR]` + specific description + the specific passage it points to
3. For CRITICAL/MAJOR issues, explain the consequence if they are not fixed

Finally, give an overall assessment:
- **Withstood challenge**: no fatal issue found in any of the five areas
- **Issues requiring attention**: issues at MAJOR level or above were found
- **Unreliable**: a CRITICAL issue was found; recommend not using it for now

---

## Material Under Review

[Paste the material under review here]

*English translation: GPT-5.5 (via Codex CLI) · 2026-07-01*