Goal Mode Changes Everything for AI Coding

Claude Code and Codex CLI have both added a /goal feature that enables autonomous looping agents — you set a high-level objective and the tool keeps iterating until it's achieved, without requiring human intervention. Practical use cases include getting GitHub Actions CI pipelines passing, running full codebase refactoring audits, and enforcing code quality checklists. The feature eliminates the need to babysit AI tools that stop early and require manual continuation. Codex has it as an experimental opt-in via /experimental settings, while Claude Code has it built in. Supporting commands include goal/pause, goal/resume, and goal/clear.