Skill UI

A set of behavioral guidelines derived from Andrej Karpathy's observations to help users writing, reviewing, or refactoring code with Large Language Models (LLMs). It guides developers to avoid overcomplication, make surgical changes, explicitly surface assumptions, and define verifiable success criteria, promoting clean, simple, and robust code development.

This skill guides the development of EarLLM One, a complex Android application. The app captures real-time audio from Bluetooth earbuds, transcribes the voice using STT, sends the text input to a Large Language Model (LLM) via a robust pipeline, and plays the response back using TTS. It requires deep knowledge of Kotlin, Android development, Bluetooth profiles (A2DP/HFP), and advanced audio routing.

A specialized module containing Yann LeCun's comprehensive arguments on the limitations of Large Language Models (LLMs). It critically examines the concept of 'emergent reasoning,' contrasts LLMs' statistical prediction with genuine world models, and addresses deep technical rivalries (e.g., LeCun vs. Hinton). Ideal for advanced discussions on AI causality, common sense, and AGI feasibility.

An expert guide for deploying and optimizing Large Language Models (LLMs) entirely on local hardware. Mastering the entire local AI ecosystem, including inference engines like Ollama, vLLM, and llama.cpp. Provides deep knowledge on quantization formats (GGUF, EXL2, AWQ), VRAM calculation, and ensuring privacy-first, offline deployment for state-of-the-art models.

This skill provides methods for reliably extracting typed, validated, and structured data from Large Language Models (LLMs), moving beyond simple text parsing. It covers best practices across major APIs (OpenAI's response_format, Anthropic's tool_use, Google's responseSchema) and local models. Key workflows include defining schemas (Pydantic/Zod), implementing constrained decoding, and building robust retry loops for production-grade data pipelines.

This tool is designed to help optimize and refine prompts used with Large Language Models (LLMs). By structuring inputs, it ensures that the instructions given to AI systems are clear, comprehensive, and contextually rich, resulting in significantly higher quality and relevance of the generated output. Useful for developers, writers, and researchers.

PydanticAI is a robust Python framework that integrates Pydantic's type-checking and validation capabilities into Large Language Models (LLMs). It allows developers to build production-ready AI agents that guarantee structured, validated outputs, moving beyond simple raw strings. Key features include supported tool calling, dependency injection for testability, and multi-model compatibility across major providers like OpenAI, Anthropic, and Gemini. Use this when reliability and correct data structure are paramount.

This skill provides comprehensive security and monitoring for Large Language Model (LLM) powered trading agents. It helps to audit trade decisions, ensure compliance with risk parameters, and mitigate potential risks associated with automated financial operations.

This skill provides comprehensive management capabilities for AI Integrations within the Arize platform. Users can create, list, retrieve, and update credentials for various Large Language Model (LLM) providers, including OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, and Vertex AI. These integrations are crucial for enabling Arize's advanced monitoring and evaluation features that require invoking external LLMs.

This skill provides a comprehensive framework for optimizing, improving, and debugging Large Language Model (LLM) prompts. It leverages production trace data, system evaluations, and human annotations to extract prompts from spans. Users can run a data-driven optimization loop, ensuring the AI responses meet desired quality standards. Ideal for prompt tuning, system prompt refinement, and improving overall AI output quality.

This skill guides the end-to-end process of building automated evaluation pipelines for Python applications powered by Large Language Models (LLMs). It teaches how to define evaluation criteria, instrument the application, build golden datasets, and run real-world evaluations using tools like pixie, ensuring the app's logic (not just the LLM) is rigorously tested. Ideal for QA, benchmarking, and improving LLM-based services.

Hypogenic is an advanced framework that uses Large Language Models (LLMs) to systematically generate and test hypotheses on complex tabular datasets. It supports three powerful modes: data-driven generation (HypoGeniC), literature-grounded refinement (HypoRefine), and combined union methods. This tool is essential for accelerating scientific discovery in fields such as predictive modeling, content analysis, and deception detection by combining empirical data insights with theoretical knowledge.