Skill UI

Before You Build acts as a crucial checkpoint in the AI coding workflow. It prevents premature implementation by rigorously assessing product risk—analyzing market demand, existing user workflows, switching costs, and distribution channels. Instead of writing code, it guides the user to define core assumptions and propose the smallest, most effective validation test (e.g., landing page test, user interview) before committing resources.

A powerful plugin enabling Claude Code to delegate complex tasks to external AI CLIs (such as OpenAI Codex or Google Antigravity/Gemini) without exiting the conversation. It supports multi-model workflows, allowing users to seek a second opinion, cross-check deep analysis, or delegate side tasks (like research or code review). The core value is that Claude synthesizes the raw outputs, providing a critical review and reconciliation of findings across all consulted models.

This skill functions as an optional agent companion for MDPR, a deterministic Markdown-to-presentation runtime. It helps review presentation workflows by providing semantic hints, visual evidence reports, and high-level layout intentions. Crucially, it keeps agent output evidence-based while letting MDPR maintain control over final slide geometry, rendering details, and layout coordinates.

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 This skill converts informal, outdated, or missing documentation into clear, concise, and professional standards without altering the core logic or behavior.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 comprehensive guide covering standards for building highly maintainable Flutter applications using Dart. It details best practices across architecture (MVVM, layered design), state management (Cubit/BLoC), folder structure, widget implementation, and testing strategies. Use this skill when writing new code, reviewing existing codebases, or planning complex features to ensure scalable and clean development.

This skill analyzes React, Tailwind, or HTML interfaces against a comprehensive design rubric (74 rules). It scores the UI from 0-100, identifying common flaws that make interfaces look generic, unfinished, or "AI-generated"—such as mixed corner radii, poor color discipline, or missing states. It provides a prioritized list of actionable fixes to ensure a polished, professional, and user-centric experience before deployment.

Transform raw structured data (CSV, Excel, JSON) into professional, long-form, interactive visualization reports. This tool generates comprehensive business review pages featuring KPI dashboards, multiple dynamic charts (using Chart.js/ECharts), detailed data tables, and narrative business insights, making it ideal for weekly or monthly performance reporting.

A specialized presentation deck template adopting a cyberpunk, CRT terminal aesthetic. It is ideal for conducting deep technical reviews, such as benchmarking CLI agents, demonstrating code diffs, or presenting complex development tool analyses. Features a dark background, grid lines, command prompt indicators, and structured tagging for maximum professional impact.

A structured presentation deck designed for rigorous review of safety incidents, security risks, and policy compliance. It utilizes distinct visual cues, including hazard stripes, L1/L2/L3 severity tiers, and dedicated code blocks for policy review (e.g., YAML). Ideal for red team exercises, post-incident reviews, and high-stakes technical briefings.

A highly visual Kanban board template designed for agile project and workflow management, similar to Trello. It features four distinct columns (To Do, In Progress, In Review, Done) to track the lifecycle of tasks. Users can filter by assignee, label, and search, and optionally group cards using swimlanes for enhanced clarity during operational sprints and team coordination.

Execute discrete, bounded coding, review, or verification tasks using the Codex CLI as an isolated subagent. This tool is ideal for tasks requiring explicit file ownership, such as isolated refactoring, feature implementation, or independent code verification, without relying on continuous conversational context. It runs in a safe sandbox environment.

This guide provides a comprehensive reference on Agent Skills, a standardized and powerful method for providing structured context, workflows, and tools to Large Language Models (LLMs). It details the anatomy of the SKILL.md file, the three-stage progressive disclosure loading process, and best practices for creating both capability (tool-based) and process (workflow-based) primitives. Use this resource when designing, reviewing, or debugging agent-driven applications to ensure reliable and context-efficient performance.