LLM & Agents

Vercel AI SDK v6 patterns for streaming chat, structured outputs, agents, and tool workflows. Load when building AI-powered features with Next.js, React, or Node.js.

分析当前会话并持久化学习到三层存储（项目 CLAUDE.md、用户 CLAUDE.md、Memory MCP）

Analyze conversation to identify and capture reusable patterns as skills, commands, agents, or hooks. Use when spotting repeatable workflows, orchestration sequences, or decision heuristics worth codifying. Triggers on pattern(s), capture, codify, reusable, workflow, or `--patternify`.

Comprehensive guidance on integrating Tailwind CSS and JavaScript in Hyvä Themes, including configuration merging, module registration, and build processes for Magento 2.

Six-layer governance hierarchy (Vision → Principles → Conventions → Development → Agents → Workflows). Use when understanding repository structure, tracing rules to foundational values, explaining architectural decisions, or navigating layer relationships.

Expert guidance for agents to manage local Nix profiles for installing tools and dependencies. Covers flakes, profile management, package searching, and registry configuration.

Three-stage content quality workflow pattern (Maker creates, Checker validates, Fixer remediates). Use when working with content quality workflows, validation processes, audit reports, or implementing maker/checker/fixer agent roles.

Fetch and download images from the internet in various formats (JPG, PNG, GIF, WebP, BMP, SVG, etc.). Use when users ask to download images, fetch images from URLs, save images from the web, or get images for embedding in documents or chats. Supports single and batch downloads with automatic format detection.

调用 OpenAI Codex 获取第二意见，批判性评估后三分类展示给用户

Consult external LLMs (Gemini, OpenAI/Codex, Qwen) for second opinions, alternative plans, independent reviews, or delegated tasks. Use when a user asks for another model's perspective, wants to compare answers, or requests delegating a subtask to Gemini/Codex/Qwen.

This skill installs a complete Claude Code skill tracking system using hooks to automatically log skill usage, execution duration, token usage, and user prompts for later analysis. Use this skill when setting up activity tracking in a Claude Code project to identify patterns, monitor costs, and discover opportunities for new skills.

Coordinate multiple AI agents working concurrently in the same workspace using GitButler's virtual branch model. Use for parallel development, sequential handoffs, and commit transfer patterns without checkout overhead.

Trunk Based Development workflow - all development on main branch with small frequent commits, minimal branching, and continuous integration. Covers when branches are justified (exceptional cases only), commit patterns, feature flag usage for incomplete work, environment branch rules (deployment only), and AI agent default behavior (assume main). Essential for understanding repository git workflow and preventing unnecessary branch proliferation

Knowledge base for the Cadre DevKit. Use when answering questions about the devkit structure, commands, skills, hooks, agents, or workflows.

创建或优化 Claude Code Skill 文件，use when creating/updating/refactoring/optimizing skill files (project)

Use when creating new skills, editing existing skills, or verifying skills work before deployment - applies TDD to process documentation by testing with subagents before writing, iterating until bulletproof against rationalization

LLM specialist router to prompt engineering, fine-tuning, RAG, evaluation, and safety skills.

When an application needs to store config, data, cache, or state files. When designing where user-specific files should live. When code writes to ~/.appname or hardcoded home paths. When implementing cross-platform file storage with platformdirs.

Use this skill when creating automated multi-phase workflows for development tasks using claude-flow orchestration. Implements patterns for implementation kickoff, verification, retry loops, and failure recovery.

Use when analyzing existing codebases to generate architecture documentation - coordinates subagent-driven exploration with mandatory workspace structure, validation gates, and pressure-resistant workflows