LLM 與 Agent

Assistente de IA local usando Ollama com modelo Gemma2 para tarefas gerais de desenvolvimento e análise

Run and evaluate the Skillset smoke test (aliases, sets, and skill invocation) and interpret its JSON reports. Use this when validating $skill/$set behavior or checking Claude Code/Codex headless runs.

Strict file creation rules. Loaded FIRST by orchestrator and all agents before any action. Prevents pollution with .md, .json, scripts. Only allows code files and .build/ docs.

Orchestrates test execution for the Nomos monorepo following TESTING_GUIDE.md standards. Use this when running tests, debugging test failures, validating coverage, or executing verification checklists from AGENTS.md files.

Generate pytest tests for Python modules with parametrization, shared fixtures in conftest.py, and minimal mocking. Use when creating unit tests, test suites, or improving test coverage. Follows 1-1 file correspondence and real object testing principles.

Use PROACTIVELY when users need guaranteed schema compliance or validated tool inputs from Anthropic's structured outputs feature. Expert advisor for choosing between JSON outputs (data extraction/formatting) and strict tool use (agentic workflows). Analyzes requirements, explains trade-offs, and delegates to specialized implementation skills. Not for simple text responses or unstructured outputs.

World-class prompt powerhouse that generates production-ready mega-prompts for any role, industry, and task through intelligent 7-question flow, 69 comprehensive presets across 15 professional domains (technical, business, creative, legal, finance, HR, design, customer, executive, manufacturing, R&D, regulatory, specialized-technical, research, creative-media), multiple output formats (XML/Claude/ChatGPT/Gemini), quality validation gates, and contextual best practices from OpenAI/Anthropic/Google. Supports both core and advanced modes with testing scenarios and prompt variations.

Technical architect assistant that helps design robust, scalable, and maintainable backend/frontend architectures. Provides visual diagrams, pattern recommendations, API design guidance, and stack selection advice. Use when designing system architecture, choosing tech stacks, planning scalability, designing APIs, or creating architectural documentation. Covers microservices, monoliths, serverless, event-driven patterns, and modern frameworks like Next.js and Supabase.

Test-driven development guidance for Tetris game logic. Use when writing tests, fixing test failures, improving test coverage, or implementing game mechanics. Auto-triggers on phrases like "write a test", "test this function", "improve coverage", "fix failing tests", or "add game logic". Emphasizes Result<T,E> pattern and property-based testing with fast-check.

Build AI-first applications with RAG pipelines, embeddings, vector databases, agentic workflows, and LLM integration. Master prompt engineering, function calling, streaming responses, and cost optimization for 2025+ AI development. Includes local LLM inference with Ollama for 93% CI cost reduction.

Capture content from JavaScript-rendered pages, login-protected sites, and multi-page documentation using Playwright MCP tools or Claude Chrome extension. Use when WebFetch fails on SPAs, dynamic content, or auth-required pages.

Claude Code 작업 결과를 6개 영역(분석, 계획, 구현, 리뷰, 검증, 테스트)에서 평가하고 구체적인 피드백을 생성합니다. 결과 수집, 영역별 점수화, P0/P1/P2 우선순위 피드백, 자기비판, 성찰, 재평가 사이클을 통해 품질을 개선합니다. Use when evaluating code results, generating feedback, reviewing implementation quality, assessing test coverage.

Document Q&A with RAG using Supabase pgvector store.

Chrome's built-in Prompt API implementation guide. Use Gemini Nano locally in browser for AI features - session management, multimodal input, structured output, streaming, Chrome Extensions. Reference for all Prompt API development.

Expert guidance for leveraging Tailwind CSS V4's new Oxide engine, CSS-first configuration, and modern styling paradigms. This skill transforms Claude into a Tailwind V4 architecture specialist, capable of designing component systems, optimizing performance, and executing complex styling challenges with precision.

Manage dotfiles with chezmoi. Use when working with dotfiles, chezmoi templates, machine-specific configuration, or when the user mentions chezmoi.

Master orchestrator control for PR lifecycle management with multi-agent coordination

Comprehensive AI development stack orchestrator for analyzing, recommending, and creating Claude Code components. Use when: (1) Planning a new project and need tech stack recommendations, (2) Analyzing existing project requirements to suggest agents/commands/hooks/settings/MCPs/skills, (3) Creating custom Claude Code components of ANY type, (4) Evaluating effectiveness of tech stacks before building, (5) Building installation command sequences from component selections, (6) User says "orchestrate", "tech stack", "what components do I need", "create agent/hook/command/setting/skill/mcp", "analyze this project", or "recommend tools". This skill combines the capabilities of skill-creator and mcp-builder into a unified orchestration layer for the entire Claude Code ecosystem.

Review and improve repository documentation including both human-readable docs (`docs/`), `README.md`, and AI agent memory context files (`CLAUDE.md`, `CODEX.md`, `AGENTS.md`, `.cursorrules`, etc.) for clarity, minimal duplication, and modularity. Use when asked to review, audit, refactor, or improve documentation structure, consolidate rules, reduce redundancy, establish shared standards, or modularize monolithic instruction files.

Automatically run the test suite when code changes are made to ensure all tests pass before proceeding. This skill runs pytest for backend projects and Playwright for frontend projects, and reports test results, coverage, and any failures.