Skill UI

An adversarial Quality Assurance (QA) pass designed to deep-dive into any codebase for logic errors, security flaws, and functional bugs. Instead of confirming that code 'looks right,' this skill actively attempts to break the code by simulating hostile inputs, tracing critical data paths, and checking for subtle, real-world correctness issues. Essential for high-stakes, critical code reviews.

This skill provides expert guidance on design engineering, focusing on UI polish, micro-interactions, and high-craft component decisions. It is ideal for reviewing frontend code (React, CSS, Tailwind) to elevate perceived quality, addressing details like easing, duration, and physical feedback. Use it when subtle detail and refined taste are critical differentiators.

This specialized skill reviews user interface animations, motion patterns, and interactive code (CSS, Framer Motion, etc.) against a strict, high-craft standard. It evaluates not only functionality but also performance, accessibility, physical correctness, and adherence to motion design principles. Use it when reviewing diffs related to transitions, keyframes, or gestures to ensure the motion feels right, not just that it runs.

This skill structures a highly controlled, bounded development life cycle for AI agents. It mandates a structured three-phase workflow: Planning (Spec), Implementation (Build), and Verification (Review). It is essential for building complex features from scratch, especially when dealing with ambiguous requirements, high risk, or needing explicit human approval gates to ensure the resulting code meets all defined specifications.

This comprehensive tool performs an architectural audit of a codebase, mapping module dependencies, checking layering integrity, and flagging structural decay. Drawing on classic engineering principles (like Conway's Law and Testability Seams), it provides a deep assessment of the project's health and structure. Ideal for large-scale code reviews or onboarding new developers.

This skill provides comprehensive pull request (PR) code reviews that go beyond simple syntax checks. It aggressively surfaces deep-rooted decay risks, architectural design smells, and maintainability issues. Findings are reported using a structured Symptom → Source → Consequence → Remedy framework, drawing insights from classic engineering principles. Ideal for reviewing complex or crucial code modules.

This skill provides a comprehensive test quality review, drawing on established principles from classic engineering texts like xUnit Test Patterns and How Google Tests Software. It diagnoses structural flaws—such as brittleness, decay, and poor coverage—in existing test suites, helping ensure that the codebase and its tests are robust, maintainable, and adhere to high industry standards.

A comprehensive guide for developers building AI demos and GPU compute tasks on Hugging Face Spaces using ZeroGPU hardware. It covers best practices for writing and reviewing code that utilizes the `@spaces.GPU` decorator, managing runtime duration and quota, and understanding the specific CUDA availability model and process isolation rules required for reliable ML deployment.

This skill provides an autonomous, repository-wide pipeline for auditing and fixing complex logical inconsistencies. It systematically executes stages including health checks, deep review, locating divergences, generating fixes, and performing diff-verification until the codebase is deemed clean. Ideal for maintaining high code quality and resolving deep, systemic bugs across entire projects.

This skill is designed to identify subtle and complex logic bugs within a single code file or function. It employs a semi-formal execution tracing methodology, systematically breaking down the suspected bug into five stages: Premises (expected state), Trace (execution path), Divergence (actual deviation), Trigger (reproduction case), and Remedy (fix). It is ideal for scenarios where users suspect code is wrong but cannot specify a concrete failure point.

This advanced skill performs code review by simulating the documented engineering philosophies of historical figures (e.g., Linus Torvalds, Steve Jobs, Ken Thompson). It moves beyond generic AI findings by grounding critiques in specific, citable principles—such as 'good taste' or 'customer experience.' Ideal for high-stakes PRs, architectural reviews, and ensuring deep, pre-submission code hardening.

This skill provides comprehensive code review capabilities for analyzing git diffs or specific file scopes. It goes beyond simple syntax checks, assessing code for reuse opportunities, quality issues (e.g., redundant state, leaky abstractions), efficiency bottlenecks, and adherence to standards. It utilizes multiple specialized sub-agents (Reuse, Quality, Efficiency, Clarity) to perform parallel, read-only audits. Optionally, it can apply high-confidence, behavior-preserving fixes upon user request.