self-learning

The self-learning skill is an autonomous system designed to research and learn new technologies directly from web sources and convert that knowledge into reusable, structured skills. It addresses the challenge of manually gathering, interpreting, and organizing technical documentation when working with unfamiliar libraries, frameworks, or tools. Instead of relying on static knowledge or prebuilt references, this skill dynamically explores the web to build up-to-date, source-grounded understanding of a given topic and transforms it into a standardized skill package.

The core capabilities of this skill include automated web discovery, selective source prioritization, and structured content extraction. It performs targeted searches such as official documentation, getting-started guides, API references, and GitHub repositories, then filters down to a small set of high-quality sources. From each source, it extracts key technical sections like installation steps, core concepts, API usage, and examples while ignoring irrelevant content such as navigation elements or ads. If standard extraction fails, it can fall back to a browser-based subagent approach for JavaScript-heavy sites. The system also normalizes topic naming into kebab-case, organizes outputs into a consistent directory structure, and ensures skills are saved either locally or globally depending on configuration.

This skill is intended for developers, AI agents, and technical writers who frequently encounter new technologies and need a reliable way to quickly understand and operationalize them. It is particularly useful in automation pipelines, developer onboarding, documentation generation workflows, and AI-assisted coding environments. By enforcing strict rules against hallucination and requiring source-based synthesis, it ensures that generated skills remain accurate, traceable, and reusable across different projects.