Skill UI

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.

This skill performs a comprehensive, read-only, multi-agent review of a git diff or file scope. It launches four specialized sub-agents—covering Regression, Security, Performance, and Contracts—in parallel. The main agent then aggregates, filters, and summarizes critical findings, ensuring the code change is robust, secure, and maintainable. Ideal for pre-commit or Pull Request review workflows.

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.'

Rewrites overly formal or vague code review comments generated by AI or corporate tools. It cuts out filler phrases like "I noticed..." or "It might be worth considering...", transforming them into direct, actionable feedback. The output uses a clear format (Location: Severity: Observation. Fix.) that sounds like it came from a helpful, experienced teammate, significantly improving the quality and reception of PR feedback.

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.

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.

Standards for maintainable Flutter/Dart apps: layer separation, feature-based structure, immutable state, modern Dart, widget choices, testing, review checklists, monorepo and localization. Use when planning, writing, reviewing, or refactoring Flutter code.

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.