Coding Agent

An AI system that pursues a software goal with its own hands: it reads and edits a codebase, runs commands and tests, observes the results, and iterates until the goal is met or a stopping rule fires. Distinct from code completion (which predicts the next lines) and chat assistance (which advises but cannot act). Examples in 2026 include Claude Code, Codex, OpenCode, Antigravity, Grok Build, Kimi Code, and IBM Bob; because they operate a real computer, they double as general-purpose workbenches for non-coding work (see Chapters 15 and 16).