Explorar skills

Complete guide for PyTorch Geometric (PyG) - graph neural networks, message passing, large-scale distributed graph learning, Lightning integration, and heterogeneous graphs

Initialize a new ML research project using the ML Research template with PyTorch Lightning, Hydra, and modern Python tooling. Use when starting a new ML project from scratch.

当用户需要根据指定时间输入（时间点、时间段或相对时间）或按最新顺序检索特定主题的新闻，并将结果整理为结构化新闻简报时使用

This skill should be used when the user asks to "research what winning means", "figure out what success looks like", "research before starting", "how to win at X", "what does winning look like for X", or wants deep research before deploying strategies. Deploys 3 research agents across 3 rounds to build a comprehensive Winning Brief.

Profile ML training performance to identify bottlenecks (data loading, compute, memory usage) and optimize GPU utilization. Use when training is slow, GPU utilization is low, or experiencing memory issues.

Run comprehensive code quality checks with ruff (format, lint) and ty (type checking). Use when checking code quality, fixing linting errors, or ensuring code follows best practices before commits or PRs.

Comprehensive guide for PyTorch Lightning - LightningModule, Trainer, distributed training, PyTorch 2.0 torch.compile integration, Lightning Fabric, and production best practices

Debug common ML training issues (NaN loss, OOM, slow training, convergence problems) and provide solutions. Use when training fails, metrics don't improve, or encountering errors like NaN loss, CUDA OOM, or slow convergence.

Use when you need to investigate an unfamiliar library, framework, architectural question, or technical decision — spawns the researcher agent which explores the codebase and searches external sources in parallel. Invoke before making technology choices or when the right approach is unclear.

Evidence-backed market sizing for bootstrapped founders. Use when user runs `/market-size`, asks to "size the market", "how big is the market", "TAM SAM SOM", "market opportunity", "is the market big enough", or needs bottom-up market research with demand validation and revenue ceiling analysis.

Capture full-page screenshots of one or more websites. Handles scroll-triggered animations, lazy-loaded content, cookie banners, and bot-protected sites. Uses a tiered approach: Microlink API first, Chrome browser fallback for protected sites. Use when you need design references, competitive analysis screenshots, or visual audits. Invoke with /site-capture:capture.

Weekly strategic intelligence. Dispatches 5 analyst agents (Market Scout, Competitor Tracker, Audience Analyst, Growth Analyst, Product Scout), reads the last 7 daily reports, and produces a strategy brief + recommendations for PM ingestion. Reads pulse-config.yaml from the nearest research directory. Run Monday mornings or whenever you need strategic direction. Trigger: "run weekly strategy", "we