Your AI coding agent just ripped through five tasks in a row. The output looks great. But did it actually commit anything? In this episode, we break down a three-layer system for keeping git hygiene built into your AI workflow — not bolted on as an afterthought. We cover standing project-level instructions, per-session git state checks, and periodic verification of what's actually in the log. Then we dive into the recovery playbook: how to handle uncommitted changes from two weeks ago, why "git add -p" beats the nuclear option, and why tagging is the cheapest insurance policy in git for solo developers working with agents. Plus, why annotated tags are better than branches when you're working solo on a single branch.
Episode #685329 — open it directly at myweirdprompts.com/685329