Adversarial Prompting

**Quick judgment** A high-quality, heavyweight analysis skill optimized for complex, high-stakes problems where you explicitly want multi-option exploration, adversarial critique, and structured consolidation. It is overkill for simple or time-urgent tasks but very strong when you need deep reasoning with visible trade-offs. **Strengths** - Enforces a disciplined 7-phase workflow (generation → critique → fixes → validation → consolidation → ranking → recommendation), which tends to improve solution quality and robustness. - Strong emphasis on surfacing failure modes, edge cases, and unintended consequences, which directly supports safer and more reliable decisions. - Produces transparent, audit-friendly reasoning that can be shared with stakeholders or used as design-review documentation. **Key risks / limitations** - Very resource- and token-intensive: multiple solutions, critiques, and fixes can lead to long runtimes and high cost, especially for large problems. - High cognitive load for users: outputs will be long, and decision-makers must sift through detailed analysis to extract what matters. - Not suitable for situations requiring rapid answers or where experimentation/iteration is more valuable than upfront analysis. - The requirement to "export the complete output to a markdown file using scripts/export_analysis.py" assumes external scripting support that may not exist in all environments; this step may be ignored or need adaptation. **Recommended scenarios** Use this skill for architecture and design reviews, complex debugging, risk analysis for strategic decisions, and any safety- or reliability-critical context where you can afford more time and compute in exchange for a deeper, adversarially-tested solution. Avoid it for quick Q&A, trivial tasks, or highly exploratory work where you intentionally do not want to pre-commit via heavy upfront analysis.