Claude Skills by frank-luongt

Automate Linear operations through Composio's Linear toolkit via Rube MCP.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: litgpt description: LitGPT framework for training, fine-tuning, and deploying LLMs with Lightning. Use when building or customizing LLM architectures with PyTorch Lightning integration. tags: [litgpt, model-architecture, training] ---

Pure C/C++ LLM inference with minimal dependencies, optimized for CPUs and non-NVIDIA hardware.

Comprehensive assistance with llama-factory development, generated from official documentation.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: llamaguard description: LlamaGuard safety classifier for LLM inputs and outputs. Use when implementing content safety filtering in production AI applications. tags: [llamaguard, safety, alignment, content-safety] ---

The leading framework for connecting LLMs with your data.

Open-source vision-language model for conversational image understanding.

> Production-ready patterns for building LLM applications, inspired by [Dify](https://github.com/langgenius/dify) and industry best practices.

You are an AI assistant development expert specializing in creating intelligent conversational interfaces, chatbots, and AI-powered applications. Design comprehensive AI assistant solutions with natural language understanding, context management, and seamless integrations.

You are an expert LangChain agent developer specializing in production-grade AI systems using LangChain 0.1+ and LangGraph.

You are an expert prompt engineer specializing in crafting effective prompts for LLMs through advanced techniques including constitutional AI, chain-of-thought reasoning, and model-specific optimization.

Master comprehensive evaluation strategies for LLM applications, from automated metrics to human evaluation and A/B testing.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: lm-evaluation-harness description: EleutherAI LM Evaluation Harness for standardized LLM benchmarking. Use when evaluating models across 200+ tasks including MMLU, GSM8K, and custom benchmarks. tags: [lm-evaluation-harness, evaluation, benchmarking] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: long-context description: Long-context extension methods for transformers. Use when extending model context windows with RoPE scaling, ALiBi, or fine-tuning approaches. tags: [long-context, emerging, research, context-window] ---

Build enterprise agents for Microsoft 365, Teams, and Copilot Studio using the Microsoft Agents SDK with aiohttp hosting, AgentApplication routing, streaming responses, and MSAL-based authentication.

Build enterprise agents for Microsoft 365, Teams, and Copilot Studio using the Microsoft 365 Agents SDK with Express hosting, AgentApplication routing, streaming responses, and Copilot Studio client integrations.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mamba description: Mamba selective state-space model architecture. Use when exploring alternatives to Transformer attention for efficient sequence modeling. tags: [mamba, model-architecture, training] ---

Comprehensive market sizing methodologies for calculating Total Addressable Market (TAM), Serviceable Available Market (SAM), and Serviceable Obtainable Market (SOM) for startup opportunities.

**(Applied - Ethical - Prioritized)** You are a **marketing psychology operator**, not a theorist. Your role is to **select, evaluate, and apply** psychological principles that: * Increase clarity * Reduce friction * Improve decision-making * Influence behavior **ethically** You do **not** overwhelm users with theory. You **choose the few models that matter most** for the situation. ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mcp-builder description: 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). tags: [mcp, tooling] ---

Patterns for building MCP servers that wrap enterprise APIs, and for consuming MCP servers across different AI agent frameworks (Claude, OpenAI Agents SDK, Google ADK).

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: megatron-core description: NVIDIA Megatron-Core for large-scale model training. Use when training 10B+ parameter models with tensor, pipeline, and expert parallelism. tags: [megatron-core, distributed-training, scaling] ---

Memory provides the persistence layer that allows agents to maintain continuity across sessions and reason over accumulated knowledge. Simple agents rely entirely on context for memory, losing all state when sessions end. Sophisticated agents implement layered memory architectures that balance immediate context needs with long-term knowledge retention. The evolution from vector stores to knowledge graphs to temporal knowledge graphs represents increasing investment in structured memory for im...

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mermaid-expert description: Create Mermaid diagrams for flowcharts, sequences, ERDs, and tags: [ai, documentation] ---

Master microservices architecture patterns including service boundaries, inter-service communication, data management, and resilience patterns for building distributed systems.

Automate Microsoft Teams operations through Composio's Microsoft Teams toolkit via Rube MCP.

miles is a high-performance, enterprise-ready RL framework optimized for large-scale model post-training. Built as a production fork of slime, it addresses critical challenges in MoE training stability, low-precision training, and train-inference alignment.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mitre-attck-reference description: MITRE ATT&CK Enterprise framework reference for mapping adversary tactics, techniques, and detection guidance tags: [security, threat-intelligence] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: ml-engineer description: Build production ML systems with PyTorch 2.x, TensorFlow, and tags: [ai, ml] ---

Expert-level guidance for writing publication-ready papers targeting **NeurIPS, ICML, ICLR, ACL, AAAI, COLM** (ML/AI venues) and **OSDI, NSDI, ASPLOS, SOSP** (Systems venues). This skill combines writing philosophy from top researchers (Nanda, Farquhar, Karpathy, Lipton, Steinhardt) with practical tools: LaTeX templates, citation verification APIs, and conference checklists.

Complete end-to-end MLOps pipeline orchestration from data preparation through model deployment.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mlflow description: MLflow experiment tracking, model registry, and deployment platform. Use when tracking ML experiments, managing model versions, or deploying models. tags: [mlflow, mlops, experiment-tracking] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mlops-engineer description: Build comprehensive ML pipelines, experiment tracking, and model tags: [ai, ml] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: mobile-security-coder description: Expert in secure mobile coding practices specializing in input tags: [mobile, security] ---

Comprehensive guide to running ML workloads on Modal's serverless GPU cloud platform.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: model-merging description: Model merging techniques (TIES, DARE, SLERP, Task Arithmetic). Use when combining multiple fine-tuned models without additional training. tags: [model-merging, emerging, research, model-composition] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: model-pruning description: Model pruning and sparsification techniques. Use when reducing model size through structured or unstructured weight removal. tags: [model-pruning, emerging, research, compression] ---

Comprehensive guide for mastering modern JavaScript (ES6+) features, functional programming patterns, and best practices for writing clean, maintainable, and performant code.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: moe-training description: Mixture-of-Experts (MoE) training and architecture patterns. Use when building or training sparse MoE models for compute-efficient scaling. tags: [moe-training, emerging, research, sparse-models] ---

Expert in monorepo architecture, build systems, and dependency management at scale. Masters Nx, Turborepo, Bazel, and Lerna for efficient multi-project development. Use PROACTIVELY for monorepo setup, build optimization, or scaling development workflows across teams.

Build efficient, scalable monorepos that enable code sharing, consistent tooling, and atomic changes across multiple packages and applications.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: multi-agent-brainstorming description: "Structured multi-agent design review with enforced roles: designer, skeptic, constraint guardian, user advocate, and arbiter. Use when validating system designs, surfacing hidden assumptions, stress-testing architectures before implementation, or preventing premature design approval." tags: [multi-agent, brainstorming, design-review] ---

Multi-agent architectures distribute work across multiple language model instances, each with its own context window. When designed well, this distribution enables capabilities beyond single-agent limits. When designed poorly, it introduces coordination overhead that negates benefits. The critical insight is that sub-agents exist primarily to isolate context, not to anthropomorphize role division.

Expert guidance for writing Python code in n8n Code nodes. ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: nanogpt description: NanoGPT minimal GPT training implementation. Use when learning GPT internals or prototyping small-scale language models from scratch. tags: [nanogpt, model-architecture, training] ---

NVIDIA's toolkit for preparing high-quality training data for LLMs.

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: nemo-evaluator description: NVIDIA NeMo Evaluator for comprehensive model assessment. Use when running multi-backend evaluations with custom benchmarks and adapters. tags: [nemo-evaluator, evaluation, benchmarking] ---

<!-- AUTO-GENERATED by export-skills.py — DO NOT EDIT --> --- name: nemo-guardrails description: NVIDIA NeMo Guardrails toolkit for LLM safety. Use when implementing programmable safety rails, topic control, or output validation for AI apps. tags: [nemo-guardrails, safety, alignment] ---

Apply these rules when writing or reviewing Next.js code. Maintained by Vercel — the authoritative source for Next.js patterns.

Comprehensive patterns for Next.js 14+ App Router architecture, Server Components, and modern full-stack React development.