What's this about?
swyx vented that git worktrees — increasingly used so multiple AI coding agents can each work on their own copy of a repo — balloon disk usage because each copy re-installs `node_modules` (20GB of duplication in his case). Separately, he called Anthropic's "ultracode" dynamic-workflow mode one of the most important coding innovations yet.
What exactly is a "git worktree"?
A worktree lets you check out several branches of the same repo into separate folders at once, so you (or several agents) can work in parallel without stepping on each other. The catch: each folder is a full working copy, so per-folder dependencies like JavaScript's `node_modules` get duplicated, eating disk fast.
Why is this notable?
It's a concrete friction point in the emerging multi-agent coding workflow — running many agents in parallel is great until the plumbing (disk, dependencies) doesn't scale. Expect tooling to fix this (shared dependency stores) as parallel-agent development goes mainstream.