Type: url
Status: generated
Language: ja
Summary
Slides from a talk in the AI track of Tech Challenge Party 2026 on 4 February 2026 that introduces agentic coding — software development using coding agents — starting from the basics of how an AI agent works. The deck traces three common failures (the agent builds something other than what was asked, code reported as “done” does not run, and rules are ignored or duplicate code spreads) to the agent not being given enough information to plan, a means to check its own work, or a mechanism that controls its actions, and answers them with three principles: define what to build, make self-verification systematic, and grow the rules that must be kept, enforcing the absolute ones with rule-based mechanisms such as hooks.