LLM 與 Agent

Java stack configuration - Maven, JUnit 5, Spotless, SpotBugs, JaCoCo with 96% coverage threshold

Spawn and coordinate parallel Claude Code sub-agents using git worktrees. Use this skill when you need to parallelize work across multiple independent tasks, delegate subtasks to sub-agents, or orchestrate complex multi-part implementations.

Android core components lifecycle, Activities, Fragments, Services, Intent system. Use when working with Android components or handling lifecycle.

Deletes specific PDF documents from Qdrant vector database collection. Use when user wants to remove, delete, or clean up PDF documents from the vector database, Qdrant collection, or needs to manage document versions.

Master the art of creating effective AI agent personas for Jules. Learn how to design role-specific prompts that guide Jules to execute tasks with precision, following best practices, and maintaining project standards. Use when creating new Jules personas or improving existing ones.

Improving frontend design beyond generic defaults. Use when creating UI components, landing pages, dashboards, or any visual frontend work. Keywords: "design", "UI", "UX", "styling", "CSS", "visual", "layout", "typography", "animation", "theme", "colors", "fonts". Addresses LLM distributional convergence toward safe, generic choices.

Use PROACTIVELY when checking if tests catch real bugs, assessing test suite quality, finding weak tests, or measuring mutation score. Validates test effectiveness beyond coverage metrics by introducing code mutations. Supports Stryker (JS/TS), PIT (Java), mutmut (Python). Not for projects without existing test suites.

Design aesthetics, typography, and layout. When Claude needs to produce designs that are inspired by Bauhaus, Minimalism, and Swiss movements.

Creates assessments with varied question types (MCQ, code-completion, debugging, projects) alignedto learning objectives with meaningful distractors based on common misconceptions. Activate wheneducators design quizzes, exams, or tests measuring understanding; need questions at appropriatecognitive levels (Bloom's taxonomy); want balanced cognitive distribution (60%+ non-recall); orrequire rubrics for open-ended questions. Generates MCQs with diagnostic distractors, code-writingprompts, debugging challenges, and project-based assessments targeting deep understanding.

Expert in building MCP servers using the Official Python MCP SDK. Use this when defining tools, resources, and prompts that allow AI agents to manage application state.

Generate Jest configuration files for JavaScript/TypeScript testing with coverage, mocking, and environment-specific settings. Triggers on "create jest config", "generate jest.config", "jest configuration for", "testing config".

Help decide where to put new instructions, conventions, or automation. Use when the user wants to add a new rule, pattern, or workflow but isn't sure if it should go in CLAUDE.md, a skill, or a command.

Deep dive into prompt-tsx patterns used in vscode-copilot-chat, including component lifecycle, async rendering, priority system, and token budget management

Q-learning, policy gradient, actor-critic, and multi-agent reinforcement learning.

Comprehensive PDF manipulation toolkit for extracting text and tables, creating new PDFs, merging/splitting documents, and handling forms. When Claude needs to fill in a PDF form or programmatically process, generate, or analyze PDF documents at scale.

Autonomous UI inspection using a dual channel: (1) visual screenshots via Playwright MCP tools, (2) numeric layout metrics via Puppeteer scripts. Includes server --check standardization so agents can start/stop reliably.

Guide for implementing Cloudflare R2 - S3-compatible object storage with zero egress fees. Use when implementing file storage, uploads/downloads, data migration to/from R2, configuring buckets, integrating with Workers, or working with R2 APIs and SDKs.

CLAUDE.md 메모리 파일 리팩토링 및 정리 스킬. 기존 CLAUDE.md 분석, 구조 개선, 컨텍스트 효율성 최적화를 수행합니다. 트리거: CLAUDE.md 리팩토링, CLAUDE.md 정리, 메모리 파일 개선, 프로젝트 가이드 최적화, CLAUDE.md 분석, 클로드 가이드 작성, CLAUDE.md 작성 도와줘.

プロジェクトの .claude/agents/ に新しいサブエージェントを作成する。「エージェント作成」「新しいエージェント」「エージェントを作って」「サブエージェント追加」「agent 作成」「エージェントを追加したい」「新規エージェント」などで起動。プロジェクト固有のサブエージェントファイルを生成。

Use semantic search to find relevant code and documentation when user asks about specific functionality, features, or implementation patterns. Automatically invoke when user asks "where is...", "how does... work", "find code that...", or similar conceptual queries. More powerful than grep for concept-based searches. Uses odino CLI with BGE embeddings for fully local semantic search.