Skill UI

This guide outlines a rigorous, technical framework for receiving and responding to code review feedback. It emphasizes that code review is a technical assessment, not an emotional exchange. Users learn how to verify assumptions, ask clarifying questions, professionally refute incorrect suggestions, and implement changes iteratively, ensuring code quality and architectural integrity.

This guide outlines a mandatory, evidence-based protocol for declaring any work complete, fixing a bug, or submitting a pull request. It strictly mandates that all claims of success must be supported by fresh, verifiable evidence, such as passing test outputs, successful build logs, or comprehensive checklist verification. It eliminates reliance on assumptions, gut feelings, or trusting proxy reports.

This tool enforces four rigorous coding principles, inspired by Andrej Karpathy's insights on LLM pitfalls. It guides developers to surface assumptions, prioritize simplicity, make surgical changes, and define verifiable goals. Use it when reviewing code, checking diffs, or before committing to ensure high code quality and prevent over-engineering.

A structured methodology for validating non-trivial decisions by subjecting them to an adversarial, fresh-context review. This process is crucial when correctness is paramount—such as in production, security-sensitive logic, or complex architecture—by forcing a skeptical examination that aims to disprove assumptions rather than simply approve them.

This advanced protocol forces the AI to conduct a thorough context audit before attempting any complex task. It mandates reading every relevant file, identifying architectural gaps, and providing a detailed orientation statement. It is essential for multi-file modifications, system design, and debugging to prevent acting on assumptions and maintain high code accuracy. The AI cannot proceed until all unknown variables are resolved.

This skill generates a detailed, step-by-step, interprocedural execution trace for a given piece of code and specific input scenario. It resolves name ambiguities, tracks type transitions, and highlights non-obvious behaviors (like implicit coercions or side effects). Use it when a user is confused about how code behaves, allowing for a deep dive into the actual execution path to clarify assumptions versus reality.

This skill enforces a rigorous, multi-phase coding workflow (Discuss -> Map -> Decompose -> Execute -> Verify) to ensure code is structurally grounded, complete, and free from invented APIs or assumptions. It is designed for high-stakes, complex, or multi-file engineering tasks, acting as a senior, disciplined coding partner that plans meticulously before writing code.