Claude Code and the /goal Command: Annoyingly Persistent
How the new /goal command changes Claude Code from giving up too easily to sticking with tough tasks.
2026-05-19
Loading content...
How the new /goal command changes Claude Code from giving up too easily to sticking with tough tasks.
I tried the new /goal command in Claude Code and I'm hooked. I need Claude to keep trying until a task is actually done. Some of these tasks are risky, like making a fragile build work. Others are low risk, like ensuring there is no corner radius larger than 16 points. Without /goal, Claude would just give up.
I don't have this problem with Grok Code in Cursor. That comes with its own risks. OpenClaw goes the other way, overly persistent, acting recklessly to hit the original goal. I wanted something in between. /goal is Claude Code's "annoyingly persistent" option.
Three things matter before you run it:
I've been using /goal to localize older apps into every iOS-supported language. Claude hits Google Translate first, then falls back to Sonnet for anything Google chokes on. Imperfect translations, but my apps went from a handful of locales to 70 plus.
Before /goal I spun up dozens of parallel agents for the same job. Now I write one prompt, walk away, and come back to 70 plus locales done.
Watch out for three things:
--dangerously-skip-permissions to keep things moving. You might stay cautious./goal keeps going longer, so a wrong assumption travels further.WOZ helps here. It pushes some of the token-wasting work back onto your machine, which cuts Claude usage and cost.