The Claude Code team just published how loops actually work. Here is the mental model free, and the 12 ready-to-run loop recipes that hand your busywork to an agent
The head of Claude Code stopped prompting.
“I don’t prompt Claude anymore. I have loops running that prompt Claude and figuring out what to do. My job is to write loops.”
That is Boris Cherny, and the numbers behind the shift are public now. Bun’s team used agent loops to rewrite roughly 750,000 lines from Zig to Rust in 11 days, with 99.8% of tests passing. Stripe compressed a migration inside a 50-million-line codebase into a single day. One dev shipped a $50K contract for $297 in tokens.
A loop is an agent repeating cycles of work until a stop condition is met. The skill is picking the right one, and the Claude Code team defines four:
Turn-based. You prompt, Claude works, you check. Every session you already run. You hand offthe checkby encoding verification as a skill.
Goal-based (/goal). You define done, and a second model judges every attempt to stop. You hand offthe stop condition.
Time-based (/loop, /schedule). The prompt reruns on an interval, locally or in the cloud. You hand offthe trigger.Proactive. Schedule + goal + parallel workflows, running with zero humans in the loop. You hand offthe prompt itself.
Read that list again as a ladder. Each rung hands the agent one more piece of your job, and most people stay parked on rung one.
That mental model is yours free. The gap between knowing it and running it is the exact commands, the verification skills, the stop conditions that resist gaming, and the cost guardrails, because the same primitives that shipped Bun’s rewrite have burned teams for $47,000 when left alone.
Behind the paywall:
▫️
exact copy-paste commands for engineering and non-engineering workThe 12 loop recipes,▫️
why a second model judges “done” and how to write conditions it can verifyThe /goal evaluator mechanic,▫️
the single change Boris says 2-3x’s output qualityThe verification skill template,▫️
what a loop beat costs, the $1,000-a-month cadence trap, and the model-routing leverThe cost math,▫️
turn caps, circuit breakers, and the blowup stories behind each ruleThe guardrail checklist,▫️
reward hacking, agentic laziness, and the Dumb ZoneThe failure-mode file,▫️
which loop for which job, on one screenThe decision table,▫️
from your first /goal to a proactive loop that runs while you sleepThe escalation path,
One subscription unlocks every system
This is one build in a growing library. Premium opens all of them:
▫️ [Loop engineering for coding agents](https://www.the-ai-corner.com/p/loop-engineering-coding-agents-2026?r=1krivi)
▫️ [The Claude managed agents guide](https://www.the-ai-corner.com/p/claude-managed-agents-guide-2026?r=1krivi)
Plus a fresh system every week. One loop that babysits your PRs for a month pays the subscription back in an afternoon.
The 12 recipes, the evaluator mechanic, the verification template, the cost math, the guardrails, and the decision table, in one system.
Try premium free for 7 days. Or get 50% off this week only.
Get The Loop Library below 👇 #
Keep reading with a 7-day free trial #
Subscribe to The AI Corner to keep reading this post and get 7 days of free access to the full post archives.