What problem does it solve? Winning Kaggle solutions contain proven techniques for NLP, computer vision, time series, tabular, and multimodal tasks, but that knowledge is scattered across writeups and notebooks. This Skill organizes extracted knowledge from top competition solutions into a searchable, continuously updated knowledge base. ## Core Features & Use Cases - Domain-Organized Knowledge Base: Browse extracted winning-solution analyses across NLP, CV, time series, tabular, and multimodal categories under references/knowledge/. - Detailed Top-Solution Analysis: Each competition file includes competition briefs, top-20 solution breakdowns with core techniques and implementation details, reusable code templates, and best practices. - Self-Evolving Extraction: Provide a Kaggle competition URL and the kaggle-miner agent extracts the winning solutions and adds them to the relevant category. - Use Case: Preparing for a mathematical reasoning competition, you consult the AIMO-2 knowledge file to learn the MARIO framework, three-stage CoT/TIR/GenSelect training, and AWQ quantization strategies used by top teams. ## Quick Start Ask the assistant to analyze the winning solutions from a specific Kaggle competition URL and add the extracted techniques to the knowledge base.
