New Skill Leaderboard

A set of engineering patterns for building accessible UI components, focusing on ARIA roles, keyboard navigation, and screen reader compatibility.

Rewrites technical text according to ASD-STE100 Simplified Technical English rules for aerospace and defense grade clarity.

Implements a think-act-prove orchestration method for AI agents featuring domain research and adversarial verification of outputs.

Converts Chinese story text or images into silent hand-drawn videos using a library of 20 artistic styles including ink, crayon, and watercolor.

An automated pipeline for creating narrated paper-collage videos, including script generation, voice-over, and motion animation.

Matches design projects to award programs and categories by verifying eligibility rules and comparing project data against official criteria.

Rewrites tool descriptions, system prompts, and technical reports using ASD-STE100 standards to eliminate ambiguity for AI agents.

A design-centric skill for creating cinematic websites and multi-screen product systems with automated visual quality gates.

Optimizes configurable system parameters for model training, inference, and black-box scientific evaluations using scalar objectives.

Detects and removes AI-generated design patterns, default semantic palettes, and machine-voice copy from web projects.

Installs and configures an Oxlint plugin to enforce opinionated rules against low-evidence patterns in TypeScript and JavaScript.

Designs or restructures folders, repositories, and knowledge bases into agent-walkable pipelines, libraries, context maps, and edit maps.

Automates the computation and preparation of Indian ITRs, including capital gains from brokerage statements and analysis of Form 16, AIS, and TIS.

Transforms images of handwriting into installable TTF, WOFF, and WOFF2 font files.

An evidence-first development skill that uses executable specs, coverage, and mutation testing to ensure high-assurance code reliability.

Reduces token usage across coding agent sessions through configurable rules for replies, artifacts, tests, code, context management, and delegation, with correctness guardrails.

A CLI tool for video rough cuts, B-roll placement, and the transcription, translation, and styling of multilingual subtitles.

Implements knowledge graph construction—including ontology design and entity fusion—and multi-agent task orchestration using graph-based workflows.

Fetches and analyzes arXiv papers to provide cited architectural recommendations and identify prior-art pitfalls for ML and systems design.

Builds source-backed analytical reports and reproducible studies to evaluate actions under uncertainty and tail risk for high-stakes decision making.

Creates hand-drawn black-pen illustrations, icons, and narrative sketches with a focus on stroke life and organic line energy.

Produces neutral briefings and maps public opinion on X regarding specific people, products, or events.

A prompting skill that re-expresses the most recent assistant message in plain, accessible language without adding new information.

An evidence-based decision board that organizes sources, ideas, and outputs in markdown to support system design and due diligence.

An autonomous security research agent that uses a looping methodology to perform deep-dive vulnerability hunts and exhaustive codebase audits.

A review tool that interrogates implementation plans against documentation and source evidence to identify unknown unknowns.

Generates editorial paper-collage B-roll clips from script lines using a gated pipeline of visual metaphor design, still-frame generation, and assemble-from-empty animation.

A German-language research skill that collects and synthesizes recent signals from news, social media, and academic sources for trend analysis and product comparisons.

Captures and recovers architectural decisions, rejected alternatives, and historical context to maintain codebase reasoning for humans and AI agents.

Optimizes AI training workloads on Moore Threads MUSA GPUs through profiling, benchmarking, and CUDA-to-MUSA performance migration.

Assesses AI and agent security using the SAIL V2 framework's 91-risk catalog, with workflows for gap analysis, compliance mapping, control prioritization, and vendor RFP generation.

A prompting skill for game prototyping that uses a critic-led loop and blind A/B testing to build games based on reference titles.

Manages local Markdown knowledge vaults by automating document capture, maintaining evidence links, and performing vault health repairs.

An orchestration protocol that uses plan gates, adversarial verification, and strict execution cards to improve the coding quality of non-frontier AI models.

Converts PDFs, slide decks, and visually dense reports into Markdown by aligning text extraction with the original document layout.

Analyzes FIT and KML files to generate performance reports, location-enriched data, and visual route stories for endurance sports.

An agent orchestration tool that offloads repetitive tasks to cheaper models based on live benchmarks and token limits.

Delegates coding, review, analysis, and research to Gemini sub-agents in background with fan-out, isolation, and verification workflows.

An automation skill for live hotel search, real-time rate verification, and end-to-end booking management.

Rents temporary private phone numbers via API to receive SMS verification codes for account signups and 2FA testing.