research-refine

What problem does it solve?

This Skill transforms vague research directions into concrete, elegant, and implementation-oriented method plans, ensuring they are anchored to the core problem and aware of the latest advancements.

Core Features & Use Cases

• Problem Anchoring: Locks down the essential problem, bottleneck, and constraints.
• Iterative Refinement: Uses a two-model loop (Claude + GPT-5.4) to iteratively improve the method plan based on expert review.
• Frontier Awareness: Integrates modern techniques (LLM, VLM, Diffusion, RL) naturally when they offer a clear advantage.
• Use Case: You have a general idea for improving LLM reasoning but aren't sure about the specific technical approach. This Skill will help you define a sharp, focused method, identify the core contribution, and sketch out minimal validation experiments.

Quick Start

Use the research-refine skill to refine my approach for making LLMs more robust to adversarial prompts.