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

Fusion Harness

ASecurity

Use when orchestrating multiple models as architect + builders — debate, collaborate, cost tracking, software factory. Triggers on "fusion harness", "multi-model orchestration", "architect builder pattern", "model debate collaboration", "cost-aware agent routing", "software factory agents".

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

Works with

claude codecli

Security Analysis

A96/100
mediumInstalls packages at runtime which could introduce malicious dependencies

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add majinmagros/magros.ai-skills --skill fusion-harness --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Fusion Harness?

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

Security grade badge for Fusion Harness
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/majinmagros-fusion-harness/badge)](https://www.skillsdirectory.com/skills/majinmagros-fusion-harness)

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

Download Zip
Files
SKILL.md
---
name: fusion-harness
description: Use when orchestrating multiple models as architect + builders — debate, collaborate, cost tracking, software factory. Triggers on "fusion harness", "multi-model orchestration", "architect builder pattern", "model debate collaboration", "cost-aware agent routing", "software factory agents".
metadata:
  origin: ECC
  source_docs:
    - https://github.com/indydevdan/fusion-harness (repo referenciado no vídeo)
    - https://docs.anthropic.com/en/docs/claude-code/sub-agents
    - https://docs.anthropic.com/en/docs/claude-code/hooks
  skills_used:
    - agent-harness-construction
    - engenharia-de-grafos
    - graph-engineering
    - cost-aware-llm-pipeline
    - autonomous-agent-harness
---

# Fusion Harness — Multi-Model Orchestration (Architect + Builders)

Fusion Harness V2 (IndyDevDan): **Architect + Builders**, modos **Debate** e **Collaborate**, custo/velocidade/tokens por modelo, **Software Factory** (out-loop). Scripts em `references/scripts/`, templates em `references/`.

## Quando usar (gatilhos concretos)

- "Configure multi-model orchestration com architect + builders"
- "Quero debate entre modelos (Fable, Gemini, DeepSeek) antes de decidir"
- "Preciso de collaboration workflow: architect planeja → builders executam → architect integra"
- "Rastrear custo/tokens/latência por modelo em tempo real"
- "Software factory: agendar agentes 24/7 para trabalhar sem supervisão"
- "Combine compute don't select — usar melhor modelo para cada tarefa"
- "Model stack registry com aliases (esconder nomes dos modelos)"

## Quando NÃO usar

- Workflows single-model (Claude Code padrão já resolve)
- Delegação simples de sub-agentes (`/task` nativo)
- Orquestração que não precisa de cost tracking ou model routing

## Arquitetura (resumo)

**Model Stack Registry** (`references/config/model-stack-template.yaml`): tiers S/A com aliases anti-sabotage (Rune=Fable 5, Flux=Gem 3.7, Drift=DS V4, Local=Qwen3.8) + custo/1k + roles.

**Workflows:** `/fh opinion` (1 pergunta → N opiniões + tabela custo/latência + síntese architect) · `/fh debate` (tese → 3 rounds → consensus) · `/fh collaborate` (goal → planos → execução paralela por nível → integração). Scripts: `references/scripts/fh-{opinion,debate,collaborate}.py`.

**Software Factory** (`references/scripts/factory-scheduler.py`): variantes SDLC (`references/variant-templates/`: nightly-refactor, pr-review, feature-build) agendadas (cron/webhook); humano só reviewa.

**Cost tracking** (`references/scripts/cost-tracker.py` + `references/settings-hooks.json`): hooks Pre/PostToolUse em `Task` → SQLite → `fh cost` dashboard/export CSV.

**Projeto** (`.fusion-harness/`): `config/` (model-stack, factory-schedule, variants) + `scripts/` + `state/` (decisions.md append-only, context.md, .locks) + `costs.db` + `logs/`. Slash commands em `references/claude-commands/`.

```bash
pip install -e .  # instala fh CLI
fh stack && fh opinion "como implementar X?" && fh cost
```

## Regras de Ouro (do vídeo)

| Princípio | Implementação |
|---|---|
| **Combine compute, don't select** | Stack multi-tier; architect escolhe por task |
| **Hide model names** | Aliases — previne sabotagem/competição |
| **Cost transparency** | Dashboard realtime; exit se budget exceeded |
| **Verification layer** | Architect = verifier; stop hooks determinísticos |
| **Out-loop > In-loop** | Factory agenda variantes; humano reviewa |
| **State persistence** | Markdown + append-only locks |

## Coordenador evoluído + variante por domínio (Batch YT rodada 4)

Orquestração agnóstica como produto: o coordenador também evolui, não só os builders.

- **Evolua o coordenador** — estratégias de coordenação em linguagem simples, refinadas por RL/loop de feedback (ex.: Sakana-style). Versione a estratégia como artefato, não como improviso do architect.
- **Variante por domínio** — uma variante do harness por domínio (ex.: Cyber: alta precisão; geral: custo). Registre acurácia por variante; promova a variante, não o modelo.
- **Camada de tráfego agnóstica** — middleware de roteamento (OpenRouter/Vercel/OpenCode-style) entre caller e modelos; troca de provider sem reescrever workflow. Combine com aliases anti-sabotage do registry.

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 →