How Do We Stop Vibe Coding?
Summary
The article critiques vibe coding, where developers rely on coding agents to generate code from intent, arguing that this shift introduces trust, understanding, and auditability problems. It analyzes the immature state of proposed solutions—markdown specs, skills, spec kits, OpenSpec, and Kiro—and contrasts them with more trust-focused approaches like CodeSpeak and Scryer, concluding that true progress requires intent-driven tooling and auditable workflows rather than prompt-driven promises.