---
type: "raw_signal"
id: "the-gauntlet-loop-the-prompting-method-behind-claude-72da39"
slug: "the-gauntlet-loop-the-prompting-method-behind-claude-72da39"
title: "The Gauntlet Loop: The Prompting Method Behind Claude of Duty"
description: "Matt Shumer’s article describes the Gauntlet Loop, a method for repeated agent improvement: give a lead agent a goal and a concrete quality reference, split the work into parts, and have separate critics compare results against that reference."
retrieval_nugget: "Matt Shumer’s article describes the Gauntlet Loop, a method for repeated agent improvement: give a lead agent a goal and a concrete quality reference, split the work into parts, and have separate critics compare results against that reference."
observed_at: "2026-10-10"
record_date: "2026-10-10"
date_kind: "observed_at"
why_it_matters: "Source: https://somethingbig.ai/gauntlet-loop"
novelty: "notable"
verification_level: "source-linked"
signal_type: "update"
source_platform: "somethingbig.ai"
topics: ["claude"]
entities: ["Matt Shumer","Claude Code","Step 5 Preview","OpenCode","Claude of Duty"]
related_patterns: ["goal-scoped-loops-replace-manual-continuation","verification-bandwidth-is-the-scarce-resource"]
source_url: "https://somethingbig.ai/gauntlet-loop"
source_urls: ["https://somethingbig.ai/gauntlet-loop"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "signal:the-gauntlet-loop-the-prompting-method-behind-claude-72da39"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/signals/the-gauntlet-loop-the-prompting-method-behind-claude-72da39/","markdown":"https://newruntime.com/signals/the-gauntlet-loop-the-prompting-method-behind-claude-72da39.md","json":"https://newruntime.com/signals/the-gauntlet-loop-the-prompting-method-behind-claude-72da39.json"}
---

# The Gauntlet Loop: The Prompting Method Behind Claude of Duty

## Retrieval answer

Matt Shumer’s article describes the Gauntlet Loop, a method for repeated agent improvement: give a lead agent a goal and a concrete quality reference, split the work into parts, and have separate critics compare results against that reference.

Matt Shumer’s article describes the Gauntlet Loop, a method for repeated agent improvement: give a lead agent a goal and a concrete quality reference, split the work into parts, and have separate critics compare results against that reference. When the reference is better, the critic identifies the gap and sends the work back for another iteration. The article says Shumer used Call of Duty screenshots as a visual bar for a game built with Claude Code, and published the prompt and project code.

A separate supplied post claims that Step 5 Preview was used to make a racing game in about an hour for roughly $4. The cited article does not substantiate that result, nor claims about the model’s OpenCode access, comparison with Kimi or GLM, or OpenRouter ranking. Those claims remain unverified in the supplied public source.

Source: https://somethingbig.ai/gauntlet-loop

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Claude](https://newruntime.com/topics/claude.md): Explore the claude topic hub.
- collection: [Raw Signals](https://newruntime.com/signals.json): Open every source-linked raw signal.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)