Before the AI era, the author wrote lengthy commit descriptions to ensure clarity and self-reflection, a process taking five to ten minutes. This practice helped explain not just the "what" but the crucial "why" of code changes, serving as a thinking tool to re-evaluate decisions. In the current agentic coding landscape, AI-generated commit messages often lack full context from offline tools, leading to fabricated reasoning that may not make sense later.
The author argues that writing commit messages personally remains vital; it forces reflection on AI-made changes and verifies intent. If the author cannot explain the "why," they are shipping code they do not understand, making future fixes difficult. Temporary decisions with exit criteria are particularly important, as writing them forces completion and helps future readers decide whether to keep the change.
Ultimately, the commit description acts as a thinking tool to ensure understanding before shipping.
Source: Hacker News · Summarized by HeadlinesBriefing