LLM & Agents

Generate comprehensive test suites ensuring 80%+ coverage for DevPortfolio. Use when asked to write tests, generate test suites, increase coverage, or create test cases. Generates Vitest + React Testing Library tests following AAA pattern with accessibility and i18n tests.

LLM keyring management, multi-provider support, and AI agent orchestration

Instrument LLM API calls with proper spans, tokens, and latency

Prepare detailed, professional prompts for Google Imagen 3/4 image generation. Supports character, environment, and object prompts using natural language with technical photography specifications. Extensible support for multiple art styles via reference files.

Search and retrieval strategies including semantic, hybrid, and reranking for RAG systems. Use when implementing retrieval mechanisms, optimizing search performance, comparing retrieval approaches, or when user mentions semantic search, hybrid search, reranking, BM25, or retrieval optimization.

FastAPI Python backend in api/. Covers routes, models, Supabase integration, authentication, R2 storage, and Cloudflare Workers deployment. Port 9999 for local dev. OpenAPI docs at /docs.

指定したプラグインをアンインストールする。「プラグインを削除」「〇〇をアンインストール」「プラグインを消して」「〇〇を削除して」「プラグインを外して」「〇〇を取り除いて」「プラグインをアンインストール」などで起動。claude plugin uninstall コマンドを使用してアンインストール。

Expert knowledge on creating Claude Code subagents. Use when designing or creating subagent .md files, understanding subagent structure, tool restrictions, or model selection.

AI-powered content and image generation using content-image-generation MCP with Google Imagen 3/4, Veo 2/3, Claude Sonnet, and Gemini 2.0. Use when generating marketing content, creating hero images, building blog posts, generating product descriptions, creating videos, optimizing AI prompts, estimating generation costs, or when user mentions Imagen, Veo, AI content, AI images, content generation, image generation, video generation, marketing copy, or Google AI.

Single source of truth for MetaSaver agent selection and subagent_type mapping

Integrate with Replicate AI for running models (image generation, LLMs, etc.).

Generates new Claude Code subagent configuration files from user descriptions. Use proactively when the user asks to create a new subagent or agent. Keywords include create subagent, new agent, build agent, agent architecture.

Continuously observes development patterns to identify automation opportunities. Always active in background. Detects repeated patterns, suggests skills/hooks/commands, auto-creates automations at threshold (3 occurrences).TRIGGERS - Keywords: patterns, learnings, observations, automation, repeated tasks, /learn:review, /learn:implement, /analyze-patterns, pending automations, skill suggestions, hook suggestions, command suggestions, optimization opportunities.TRIGGERS - Phrases: "analyze patterns", "review learnings", "what have you learned", "create automation", "automate this", "I keep doing this", "repeated task", "optimize workflow", "suggest improvements", "pending suggestions".TRIGGERS - Automatic: After file modifications (via hook), before git commits, at session end, when dependencies added, after task completion.

Guides on analyzing and improving test coverage for Python (pytest-cov) and JavaScript (c8) projects. Use when a user asks to set up test coverage, generate coverage reports, understand why coverage is low, or improve their test coverage.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Coil Selection & Thermal Design Agent

Use this skill for semantic code search and codebase understanding. CCG-RAG provides intelligent retrieval using code embeddings and knowledge graphs.

Manage feature labeling workflow - queue management, label storage, similar features, progress tracking

Repository type detection for identifying library vs consumer repositories. Analyzes directory structure, package.json dependencies, and monorepo indicators to classify repositories and detect multi-mono relationships. Returns metadata about workspace type, monorepo tool (Turborepo, nx, lerna, pnpm-workspace), and library consumption patterns. Use when agents need to adapt behavior based on repository type or validate architecture-specific patterns.

Best practices for writing Python tests - use when creating or improving test coverage