Skill UI

Behavioral rules for LLMs writing or reviewing code, emphasizing caution, simplicity, surgical edits, and goal-driven execution to avoid common mistakes during implementation tasks, bug fixes, or refactors.

A comprehensive guide for auditing and ensuring regulatory compliance (SOC2, GDPR, HIPAA) when integrating Large Language Models (LLMs) via OpenRouter. It covers best practices for data handling, access control, PII redaction, implementing robust audit trails, and configuring provider routing to maintain data sovereignty and meet strict industry standards.

This skill provides comprehensive reference architectures for building production-grade AI applications using OpenRouter as a unified LLM gateway. It covers three scaling patterns: simple single-service, standard microservice (with caching and budget enforcement), and enterprise event-driven systems (with message queues and workers). Use this when planning system design, reviewing complex AI pipelines, or ensuring scalability and cost control in your LLM integration.

Runs external LLM CLIs (OpenAI Codex or Google Gemini) to review uncommitted changes, branch diffs, or specific commits, providing an independent second opinion before merges or releases.

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 autonomously improves generated academic papers by simulating a multi-round review, fixing, and recompilation cycle. It is designed to enhance writing quality, fix theoretical inconsistencies, and improve presentation structure after the initial drafting phase. It operates in rounds (default 2 rounds) using external LLM review and can optionally accept a style reference paper to guide structural fixes.

Behavioral guidelines for writing, reviewing, and refactoring code with LLMs, emphasizing minimalism, clear assumptions, surgical edits, and verifiable success criteria to avoid common machine-assisted coding mistakes.

This tool enforces four rigorous coding principles, inspired by Andrej Karpathy's insights on LLM pitfalls. It guides developers to surface assumptions, prioritize simplicity, make surgical changes, and define verifiable goals. Use it when reviewing code, checking diffs, or before committing to ensure high code quality and prevent over-engineering.

MockHunter is an advanced skill designed to audit live web pages, providing a comprehensive reality check on visible data. It analyzes every element to determine its true source: whether it is REAL (from the database), MOCK (simulated data), LLM-generated, HARDCODED, BROKEN (broken endpoints), or UNKNOWN. Ideal for auditing AI-generated UIs or reviewing proof-of-concept MVPs before customer sign-off.

This skill generates comprehensive, academic-style survey papers as a single, self-contained HTML file. By reading an anchor resource and curating research data, it structures the content and utilizes advanced LLMs (like Kimi K2.6 via Fireworks) to generate all prose, figures, and bibliographies, suitable for rapid academic literature review.

This guide provides a comprehensive reference on Agent Skills, a standardized and powerful method for providing structured context, workflows, and tools to Large Language Models (LLMs). It details the anatomy of the SKILL.md file, the three-stage progressive disclosure loading process, and best practices for creating both capability (tool-based) and process (workflow-based) primitives. Use this resource when designing, reviewing, or debugging agent-driven applications to ensure reliable and context-efficient performance.