Skill UI

Empowers building AI agents that perceive screens, plan via vision-language reasoning, and control mouse/keyboard for desktop automation, with attention to sandboxing and safety across Anthropic, OpenAI, and open-source paradigms.

A comprehensive guide to understanding the complete state (context) available to Large Language Models during inference. Covers core components like System Prompts, Tool Definitions, Message History, and Retrieval Augmented Generation (RAG). Learning to engineer context is crucial for building robust, efficient, and reliable AI agent systems by managing the attention budget and ensuring high-signal information delivery.

This skill guides the development of robust, security-focused full-stack web applications. It ensures layered security coverage across the entire application stack, from database queries (parameterized queries) to API endpoints (authentication/authorization) and frontend components (input validation, output encoding). Use this when building end-to-end features, microservices, or REST APIs that require simultaneous attention to Frontend, Backend, and Security best practices.

Flash Attention optimizer delivers 2-4x speedup and 10-20x memory savings for transformer attention, ideal when training or running long-context models that hit GPU memory or latency limits; supports PyTorch native scaled_dot_product_attention, the flash-attn library, H100 FP8 precision, and sliding-window attention.

Serve production-grade LLM APIs with vLLM, leveraging PagedAttention, continuous batching, quantization, and tensor parallelism to maximize throughput while keeping latency, GPU usage, and scalability under control for OpenAI-compatible deployments.

SGLang is a high-performance serving framework for LLMs and VLMs that uses RadixAttention prefix caching to deliver structured outputs (JSON/regex/grammar) and agentic workflows with function calls, providing up to 5× faster inference than vLLM across multi-GPU production deployments.

TransformerLens is the standard library for performing mechanistic interpretability research on large language models. It provides clean interfaces to inspect and manipulate transformer internals—such as attention patterns, residual streams, and MLP outputs—via activation caching and HookPoints. This tool is essential for reverse-engineering model algorithms, performing causal tracing, and studying the internal circuits responsible for model behavior.

This skill fetches unread GitHub Pull Requests (PRs) that require review from a specified team or were opened by a team member. It is ideal for users who need to check their team's review queue, find PRs needing attention, or automate the process of reviewing open code submissions without manually browsing GitHub notifications. Requires GitHub CLI authentication.

This methodology converts verbose, natural language system instructions into an atomic, structured-label format (e.g., trigger:, action:). By structuring rules, it bypasses the limitations of natural language processing, ensuring LLMs allocate full attention to every constraint, thereby maximizing compliance, improving reliability, and reducing token usage.

A specialized cognitive psychology skill designed to engineer highly engaging headlines, subject lines, and titles. It utilizes principles like the curiosity gap, self-relevance, and contrast to create compelling hooks for marketing copy, landing pages, emails, and social media ads. It ensures attention-grabbing power while maintaining ethical guardrails and promise continuity.

Identify past-due Google Tasks via gws commands so items needing attention are surfaced quickly. The recipe lists task lists, filters for active tasks, and reviews due dates to spotlight and prioritize overdue work.

This tool searches the current GitHub repository and retrieves all issues explicitly assigned to the user. It goes beyond simple listing by analyzing key issue metadata—such as age, comment volume, and current status (open/closed)—to proactively suggest which issues require immediate attention or focus. It is indispensable for contributors, project managers, or developers who need a quick, prioritized overview of their assigned workload.