Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

Back to skills

Gauntlet Loop

ASecurity

Use when a task needs an exceptional/impressive result and you can afford many tokens and time — large creative builds (games, apps, sites, 3D scenes) with subagents. Triggers on "gauntlet loop", "gauntlet-loop", "resultado de outro nível", "impressionar", "arquitetura de agentes", "subagentes em paralelo", "julgamento às cegas". Same family as graph-engineering/score-loop but adds BLIND judgment (verifier never sees the creation context) and the "impress the supervisor" acceptance bar. Do NO...

2 stars
0 votes
0 copies
0 views
Added 9/19/2026
ai-agentsgo

Security Analysis

A100/100

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add majinmagros/magros.ai-skills --skill gauntlet-loop --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Gauntlet Loop?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Gauntlet Loop
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majinmagros-gauntlet-loop/badge)](https://www.skillsdirectory.com/skills/majinmagros-gauntlet-loop)

More formats (shields.io, HTML) on the badges page.

Download Zip
Files
SKILL.md
---
name: gauntlet-loop
description: Use when a task needs an exceptional/impressive result and you can afford many tokens and time — large creative builds (games, apps, sites, 3D scenes) with subagents. Triggers on "gauntlet loop", "gauntlet-loop", "resultado de outro nível", "impressionar", "arquitetura de agentes", "subagentes em paralelo", "julgamento às cegas". Same family as graph-engineering/score-loop but adds BLIND judgment (verifier never sees the creation context) and the "impress the supervisor" acceptance bar. Do NOT use for small/fast tasks — it burns a lot of tokens.
---

# Skill: Gauntlet-loop — segmentação + verificadores às cegas

Arquitetura para entrega de nível "uau" (inspirada na técnica Gauntlet Loop).
Quebra a tarefa em muitos segmentos, dá a cada um um par
executor + verificador, e só libera o que deixa o verificador IMPRESSIONADO.

Custo alto: horas + centenas de milhares de tokens. Usar só quando o resultado
precisa ser excepcional.

## Quando usar

- Projetos criativos grandes: jogo, app, site, cena 3D, simulador.
- Resultado precisa ser "impressionante", não apenas "funciona".
- Você pode deixar o agente trabalhando por horas.
- Tens orçamento de tokens para múltiplas rodadas + verificação.

NÃO usar para: bug fix, tarefa pequena, output que uma nota 0-100 resolve.
Nesses casos use `graph-engineering` / `score-loop` / `superpowers`.

## 2. Pipeline (4 etapas)

```
1. DEFINIR   entrada = tarefa + meta + critérios de aceitação
2. SEGMENTAR quebra o projeto em N partes pequenas
3. PARALELIZAR por segmento: subagente EXECUTOR + subagente VERIFICADOR
             verificador julga ÀS CEGAS (sem contexto) e precisa ficar
             impressionado (barra "uau") para aprovar
4. ITERAR    reprovação → volta ao loop do segmento; tudo aprovado → relatório
```

## 3. Etapa 1 — Definir (input)

Capture do usuário:

| Campo | Exemplo |
|---|---|
| **Tarefa** | "simulador de voo em duna com nave 3D" |
| **Meta** | resultado observável e desejado |
| **Critérios de aceitação** | do PROJETO (ex.: "comparar com jogos comerciais") |
| **Barra de aprovação** | por padrão: verificador IMPRESSIONADO (surpresa real) |
| **Orçamento** | rodadas máx (ex.: 8), tempo máx, tokens máx |

Se a tarefa for vaga, pergunte antes (não invente critério).

## 4. Etapa 2 — Segmentar

- Divida o projeto em **muitas partes pequenas e independentes** (ex.: cenário,
  física, nave, câmera, som, UI — cada uma é um segmento).
- Cada segmento tem saída verificável isolada.
- Considere o tamanho: tarefa complexa → mais rodadas permitidas.

## 5. Etapa 3 — Paralelizar (executor + verificador)

Para CADA segmento, crie o par:

- **EXECUTOR**: subagente que constrói o segmento sozinho.
- **VERIFICADOR**: subagente separado que avalia o resultado.

**Regras do verificador (o diferencial):**

Attribution

majinmagrosmajinmagros
View sourceMore from majinmagros →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode. Cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Supports intensity levels: lite, full (default), ultra, wenyan-lite, wenyan-full, wenyan-ultra. Use when user says "caveman mode", "talk like caveman", "use caveman", "less tokens", "be brief", or invokes /caveman. Also auto-triggers when token efficiency is requested.

1023331 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

686011 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3331 votes

catchup

Recovers prior coding-agent session context by running `catchup <agent> --since-compact`, which extracts a clean summary of a previous Codex, Claude Code, Antigravity, OpenCode, or Pi Agent session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", or asks to recover/summarize a previous session before continuing. Do NOT use for the current conversation, git history, or any non-agent log.

611 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →