As artificial intelligence (AI) coding tools spread, developers are using the concept of "harness engineering," defined by An Su-bin (안수빈), a research engineer at Asteromorph, as "giving hands and feet to AI work." The idea is to put a harness on AI and make it move as desired. AI work proceeds through four processes: brain (LLM as brain), planning, memory and tool use, a concept systematised by Lilian Weng (릴리안 웽) in her 2023 blog post "LLM Powered Autonomous Agents." The developer acts as a "foreman" who tells the LLM what to do through the LLM context window, setting fixed rules with a system prompt or limiting search range with retrieval-augmented generation (RAG). Harness engineering encompasses this design work. As know-how accumulates, it can be packaged in a reusable form called a "skill," which turns tacit knowledge into explicit knowledge. Skills spread through the platform "skills (skills.sh)" unveiled by Vercel in January, after Anthropic released the skill format as an open standard in December 2025. More recently, a "skill creator" has emerged. The direction of AI-use techniques has shifted from controlling LLM behaviour to combining models, leading to the "multi-agent" concept, with Anthropic's official pattern being "Orchestrator-Workers."