Skill UI

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.

A systematic, non-mutating investigation designed for complex software failures, including bugs, regressions, crashes, and flaky behavior. This skill launches four specialized, read-only sub-agents in parallel to perform deep root-cause analysis, trace code paths, pinpoint the smallest failing seam, and predict the quickest path to confirming or fixing the issue. It is designed purely for diagnosis and hypothesis generation.

This skill provides a comprehensive set of core UI components (e.g., Box, Button, Input, Card) and implements a structured design system. It guides developers on using design tokens (like `$textPrimary` or `$4`) instead of hard-coded values, ensuring visual consistency and scalability across applications. Ideal for any project requiring standardized, maintainable UI development.

This skill provides comprehensive best practices for designing modern and robust React user interfaces. It covers critical patterns for managing asynchronous data, including optimal loading state indicators (skeletons vs. spinners), structured error handling hierarchies, and designing contextual empty states. Essential for improving user experience during data fetching and operations.

This skill is designed to accurately convert complex, Hugo-managed documentation sites and content into standard Markdown format. It goes beyond generic knowledge by inspecting the entire local Hugo repository structure, including `hugo.toml`, custom shortcodes, layouts, and content files. It preserves the original content's semantic meaning while materializing Hugo-specific constructs into clean Markdown.

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 library provides a comprehensive framework for managing bounded AI workflows (loops). Users can analyze existing codebases or work histories to discover opportunities for repeating, automated agent processes. It guides the user through finding published templates, auditing and repairing weak agents, adapting parameters, or crafting entirely new, robust automation designs, ensuring clear triggers, verification, and mandatory stopping conditions.

StyleSeed Design Review analyzes UI/frontend code (React/Tailwind/HTML) against a detailed design rubric. It scores the interface (0-100) across coherence, color discipline, hierarchy, and polish. The tool pinpoints specific violations—such as mixed corner radii or non-semantic colors—and generates a prioritized, actionable list of fixes, ensuring the final product looks professional and polished, not generic or 'AI-generated.'

A fast, grep-based linter that checks for common design system violations in CSS and component code, such as hardcoded colors, raw pixel values, and deprecated syntax. It provides immediate automated feedback to ensure code adheres to established design tokens and best practices.

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.

A constraint-based protocol designed to professionalize existing code by normalizing comments and performing safe, non-semantic cleanup. This skill converts informal, outdated, or missing documentation into clear, concise, and professional standards without altering the core logic or behavior. Ideal for preparing code for code review, open-sourcing, or formal documentation.

A professional layout template designed for publishing long-form articles, newsletters, or essays on platforms like Substack, Medium, or WeChat Official Accounts. It mimics a high-end magazine style, featuring distinct sections for hero images, structured text blocks, styled quotes, and code snippets to enhance readability and visual appeal.