What's hiding in your Claude Code config files? I mined mine-found 22 patterns (astgl.com)

An AI developer conducted an experiment by mining their Claude Code config files, uncovering 22 distinct patterns that had emerged from months of interactions with the AI. This involved systematically reviewing over 80 files, including global instructions, project memory indices, and custom commands, to identify recurring workflows and preferences. The process entailed organizing these implicit patterns into structured runbooks, termed SKILL.md files, which transform vague instructions into explicit, reusable resources. This discovery is significant for the AI/ML community as it highlights the importance of recognizing and formalizing repetitive workflows within AI configuration files—a concept referred to as "skill mining." By making implicit knowledge explicit, developers can streamline their processes and improve efficiency across multiple projects. The findings showcased patterns in diverse domains, like development workflows and content management, thus emphasizing the interconnectedness of tasks even when they were created independently. The technique is also adaptable, applicable not only to Claude Code but to any AI tool where users accumulate a body of instructions, making it a valuable strategy for enhancing productivity and clarity in AI-assisted development.