AI & LLM

This skill provides comprehensive guidance for creating effective Claude Code skills. It includes scripts to initialize new skill directories with proper templates, validate skill structure and frontmatter, and package skills into distributable .skill files. The skill also documents best practices for skill design including progressive disclosure patterns and resource organization.

This skill provides a complete guide for building MCP (Model Context Protocol) servers that enable LLMs to interact with external services. It includes implementation patterns for Python (FastMCP) and TypeScript (MCP SDK), best practices for tool design, and an evaluation harness to test MCP server quality using Claude.

Migrate prompts and code from Claude Sonnet 4.0, Sonnet 4.5, or Opus 4.1 to Opus 4.5. Use when the user wants to update their codebase, prompts, or API calls to use Opus 4.5. Handles model string updates and prompt adjustments for known Opus 4.5 behavioral differences. Does NOT migrate Haiku 4.5.

Guides the creation of well-structured Claude Code skills by applying Test-Driven Development principles to documentation. Includes templates for SKILL.md structure, testing methodologies using subagents with pressure scenarios, and optimization techniques for skill discovery (Claude Search Optimization).

This skill provides a conceptual framework for progressively refining context retrieval in multi-agent workflows. It describes a 4-phase DISPATCH-EVALUATE-REFINE-LOOP pattern to solve the 'subagent context problem' where agents don't know what context they need until they start working.

This skill runs as a Stop hook at the end of Claude Code sessions to analyze conversation transcripts and extract reusable patterns. It identifies error resolutions, debugging techniques, workarounds, and project-specific conventions, saving them as learned skills for future reference.

Continuous Learning v2 is an advanced learning system that observes your Claude Code sessions through hooks, capturing tool usage patterns. It creates atomic 'instincts' with confidence scoring that can evolve into reusable skills, commands, and agents. The system learns your preferences and workflows to make Claude more personalized over time.

A comprehensive guide for selecting and optimizing embedding models for vector search and RAG applications. Provides code templates for Voyage AI, OpenAI, and local embedding models, along with chunking strategies, domain-specific pipelines, and quality evaluation methods.

This skill provides comprehensive guidance for building sophisticated LLM applications using LangChain 1.x and LangGraph. It covers agent orchestration with StateGraph, RAG implementations, multi-agent systems, memory management patterns, and production deployment best practices including LangSmith observability integration.

This skill teaches comprehensive evaluation strategies for LLM applications, covering automated metrics (BLEU, ROUGE, BERTScore), human evaluation frameworks, LLM-as-Judge patterns using Claude, A/B testing with statistical analysis, and regression detection. It includes ready-to-use Python code examples and integrates with tools like LangSmith.

This skill provides comprehensive documentation and code examples for building RAG (Retrieval-Augmented Generation) systems. It covers vector database setup (Pinecone, Weaviate, Chroma, pgvector), embedding strategies, retrieval patterns (hybrid search, HyDE, multi-query), chunking strategies, and evaluation metrics using LangChain and LangGraph.

A comprehensive prompt engineering skill that teaches advanced techniques like chain-of-thought, few-shot learning, structured outputs, and prompt optimization. Includes reference documentation, template libraries, and a Python script for A/B testing prompts to improve LLM performance.

A comprehensive MLOps skill that provides specialized AI agents (data scientist, ML engineer, MLOps engineer) and workflow templates for building production machine learning pipelines. It guides users through data preparation, model training, validation, deployment, and monitoring stages using modern ML tools like MLflow, Kubeflow, and Feast.

A comprehensive plugin for building production-ready LLM applications, including RAG systems with vector databases, AI agents using LangGraph, advanced prompt engineering patterns, and embedding strategies. It provides templates and best practices for integrating with Pinecone, Qdrant, pgvector, and other vector stores.

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.

Guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).

Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies

Claude Flow is an enterprise-grade AI agent orchestration plugin providing 150+ commands, 74+ specialized agents, and swarm coordination capabilities. It enables multi-agent workflows with SPARC methodology, GitHub automation, neural training, and cross-session memory persistence through MCP server integration.

A comprehensive collection of educational skills for learning context engineering principles. Covers context window management, multi-agent coordination patterns, memory system design, tool creation, and evaluation frameworks. All scripts are demonstrations using mock data to illustrate concepts without external dependencies.

This skill provides comprehensive guidance for PyTorch Geometric (PyG), a library for developing Graph Neural Networks. It covers graph creation, GNN architectures (GCN, GAT, GraphSAGE, GIN), node/graph classification, molecular property prediction, and large-scale graph learning with extensive reference documentation and utility scripts.

A comprehensive reference skill for reinforcement learning with Stable Baselines3. It provides algorithm selection guides, training templates, custom environment creation tutorials, callback documentation, and vectorized environment usage patterns for efficient RL agent development.

This skill provides comprehensive documentation and templates for PyTorch Lightning, a framework that organizes PyTorch code for scalable deep learning. It includes ready-to-use templates for LightningModules and DataModules, Trainer configurations for various scenarios (single GPU, multi-GPU, FSDP, DeepSpeed), and detailed guides for callbacks, logging, distributed training, and best practices.

A comprehensive molecular machine learning skill using DeepChem for predicting chemical properties like solubility and toxicity. It supports graph neural networks, transfer learning with pretrained models (ChemBERTa, GROVER), and MoleculeNet benchmarks for drug discovery and materials science applications.

DSPy Ruby is a comprehensive guide for building LLM-powered Ruby applications using the DSPy.rb framework. It provides type-safe signatures, composable modules, multi-provider support (OpenAI, Anthropic, Gemini, Ollama), and patterns for testing, optimization, and production monitoring of AI applications.

A detailed reference skill for designing agent-native software architectures. Provides patterns for tool design, file-based workspaces, mobile execution, self-modification with safety guardrails, and progressive disclosure of agent capabilities. Use when building apps where features are outcomes achieved by agents operating in loops.

This skill should be used when working with pre-trained transformer models for natural language processing, computer vision, audio, or multimodal tasks. Use for text generation, classification, question answering, translation, summarization, image classification, object detection, speech recognition, and fine-tuning models on custom datasets.

Create new Claude Agent SDK projects with proper setup and configuration. Choose between TypeScript or Python, get the latest SDK version, and automatically verify your application follows best practices.

This skill provides comprehensive documentation for writing hookify rules - markdown files with YAML frontmatter that define patterns to watch for and messages to show when those patterns match. It covers rule file format, event types (bash, file, stop, prompt), regex pattern syntax, advanced conditions, and best practices for organizing rule files.

This skill teaches developers how to create autonomous agents for Claude Code plugins. It provides comprehensive documentation on agent file structure, YAML frontmatter fields, system prompt design patterns, and triggering conditions with concrete examples. Includes a bash validation script to check agent files for correct structure.

MCP Agent Mail enables multiple AI coding agents to coordinate asynchronously through a mail-like system. It provides inbox/outbox messaging, file reservations to prevent conflicts, and contact management, with all communications stored in Git for human auditability.