Skill UI

Plans, audits, and safely runs ChicagoHAI HypoGeniC/HypoRefine workflows for LLM-assisted hypothesis generation from labeled text datasets, with pinned installs, config validation, dataset leakage checks, and bounded local review before any model call.

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.

Generates a comprehensive Chinese interview-prep cheat sheet for a specific ML/LLM topic, including formulas with derivations, from-scratch PyTorch code, comparison tables, and 25 high-frequency interview questions at L1/L2/L3 levels. Ideal for interview preparation and focused technical review.

This skill provides a generic entry point for delegating self-contained, complex tasks to various LLM-powered sub-agents (Codex, Claude Code, Pi). It is designed to be used in meta-skill DAG steps for tasks like feature creation, PR reviews, code refactoring, or document generation, providing full tool surface access. It favors non-interactive CLI modes for reliable background execution.

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.

A comprehensive guide for researchers and reviewers designing or auditing Automated Software Engineering (ASE) evaluations. It outlines the necessary rigor, including the use of real subject systems, establishing fair and runnable tool baselines, performing targeted ablations (especially for LLMs), defining explicit oracles, and ensuring full data and model provenance to guarantee reproducible and scientifically sound results.

A comprehensive guide for researchers submitting to ASE (Automated Software Engineering). It details mandatory requirements for ensuring research reproducibility, covering the Data Availability Statement, providing anonymized and runnable tools, pinning system provenance (commits, versions, SHAs), and managing cached LLM outputs. Essential for passing rigorous peer review.

A comprehensive guide for authors on best practices for packaging, documenting, and ensuring the reproducibility of LLM artifacts (weights, data, code, prompts, etc.). It addresses requirements for anonymous academic review and public post-acceptance release, focusing on reproducibility, licensing, and transparency.

A comprehensive guide detailing the academic review pipeline for the COLM conference. It covers the entire journey from abstract submission to final decisions, emphasizing unique elements such as LLM use rules, reciprocal reviewing obligations, and strategic methods for maximizing impact during the rebuttal phase. Essential reading for authors planning submissions to specialized, young venues.

A comprehensive model simulating the entire academic paper review cycle, specifically tailored to large-scale top-tier conferences like CVPR. It details the timeline, roles of reviewers and Area Chairs, the impact of mechanisms (like the LLM ban), and strategic advice for authors navigating the submission, rebuttal, and decision-making phases.