Skill UI

This skill guides the end-to-end process of building automated evaluation pipelines for Python applications powered by Large Language Models (LLMs). It teaches how to define evaluation criteria, instrument the application, build golden datasets, and run real-world evaluations using tools like pixie, ensuring the app's logic (not just the LLM) is rigorously tested. Ideal for QA, benchmarking, and improving LLM-based services.

Automates installing and configuring language servers for Copilot CLI, detecting the OS, merging user or repo configs, and validating binaries so go-to-definition, hover, and other code intelligence work across languages.

A comprehensive, multi-phase quality engineering audit tool designed to test codebases beyond standard unit tests. It first explores the codebase's domain and architecture to derive behavioral requirements, then executes a rigorous sequence including spec-traced functional tests, three-pass code reviews, and a multi-model specification audit (Council of Three). It is specifically engineered to find the critical 35% of real-world defects that structural code review alone might miss. Works with any programming language.

Transforms shorthand, quasi-code, or fuzzy natural-language notes into production-ready implementations, bridging collaborators with different expertise and ensuring robust, expert-quality output.

Convert a natural-language project specification into a runnable starter repository by parsing intents, planning architecture, and generating real code plus CI/env files for any tech stack.

A comprehensive tool that analyzes complex codebases to generate detailed, structured architectural blueprints. It automatically detects and documents the entire technology stack, including primary languages (.NET, Java, Python, etc.), dependencies, versions, licensing, and implementation patterns. The output provides actionable documentation covering coding conventions, usage examples (API, data access), and visual architecture maps, ensuring consistency across guided software development projects.

This guide provides comprehensive guidelines for localizing all aspects of VS Code extensions. It covers best practices for handling configurations (settings, commands, menus), user-facing string resources, and walkthrough content. Developers can use this skill to ensure their extensions are properly adapted and functional for multiple languages.

A comprehensive guide for professional code review in Chinese development teams. It addresses common communication pitfalls, offering structured methods to give critical feedback that is both rigorous and constructive. Learn how to use suggestion-based language, structured severity levels, mixed Chinese/English commenting, and optimal review processes to elevate code quality and team collaboration.

This skill is designed for understanding complex database structures. It allows users to list all available tables, intelligently filter relevant tables based on natural language queries, retrieve detailed schemas (columns, types, primary/foreign keys), and map inter-table relationships. It culminates in generating the M-Schema format, which is essential for accurately preparing the context needed for advanced SQL query generation.

A powerful utility for performing semantic searches across a large knowledge base of 105K+ skill cards. It allows users to find the most relevant skills based on natural language queries, supporting both Python API integration and Command Line Interface (CLI). This tool is zero-dependency and highly efficient for skill discovery and knowledge retrieval in development workflows.

Provides technology-agnostic guidance for architecting and reviewing modular systems. This skill covers core architectural principles such as bounded contexts, state isolation, explicit contracts, and failure containment. Use it when determining service boundaries, assessing coupling between domains, or designing new greenfield architectures without assuming a specific framework, language, or repository structure.

This guide details advanced strategies for optimizing Claude API usage to reduce latency and cost. Techniques covered include prompt caching (significant cost reduction), strategic model selection (Haiku, Sonnet, Opus) based on task complexity, implementing streaming for perceived speed, and managing token limits. Use this guide when building high-performance, cost-efficient LLM applications that interact with large language models.