convez.ai
SkillAutomaçãoAvançado
Jott2121

Fleet Mode

Jott2121/fleet-mode

Orquestração inteligente de agentes com bias para simplicidade e custo-eficiência

Ver no GitHub
InicialPluginScore · de 0 a 100

Uma doutrina medida para orquestração de agentes de IA, sabendo quando não usar múltiplos agentes. Implementa quatro regras fundamentais: leituras paralelizam, escritas são single-threaded, verificações determinísticas precedem review, e cada agente adicionado deve justificar seu custo. Roda como skill nativa do Claude Code com gates humanos para decisões irreversíveis e logging completo de decisões.

Instalação

$npx convez-ai Jott2121/fleet-mode

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
1
Acompanhando
0
Saúde
42%
Último commit
há 4 meses
Versão
—
Issues abertas
1
Contribuidores
1

Casos de uso

  • Coordenar múltiplos agentes IA em fluxos de trabalho autônomos sem paralisia por análise
  • Implementar gates de qualidade e revisão independente antes de ações irreversíveis
  • Registrar e auditoria cada decisão de agente mantida ou descartada com métricas reais
  • Reduzir custos de orquestração decidindo quando usar um agente único vs paralelismo

Veredito

Fleet Mode é um framework opcionado e pragmático que inverte a mentalidade comum de 'mais agentes = melhor'. Excelente para quem opera sistemas autônomos em produção e precisa de controle de custo real, auditoria de decisões e segurança operacional — mas requer disciplina para manter o single-threading de escritas e os gates humanos.

README

A tradução para o português deste README ainda está na fila.

Fleet Mode

A measured doctrine for orchestrating AI agents, and knowing when not to.

Fleet Mode

Most "multi-agent" advice assumes more agents is better. It usually isn't. Fleet Mode is the opposite stance, run as a live Claude Code skill: adding agents has a negative average payoff on most tasks, so you fan out only where it demonstrably earns it, and you keep the writes single-threaded and gated.

It's the operating mode behind bow (an autonomous all-Claude chief-of-staff agent) and the builds it ships.

🧩 One layer of a five-repo cost-governance stack for operating AI agents cost-efficiently; bow is the flagship that runs every layer in production.


The four rules

  1. Read-heavy work fans out. Research, codebase/PR review, multi-file audits: parallel subagents in clean contexts, each returning a condensed summary. This is the only thing fan-out is for.
  2. Writes stay single-threaded. One agent makes the edit. Never fan out to edit in parallel.
  3. Deterministic checks first, then an independent refute-first review. Tests/types/lint/build must pass with no model in the loop; then a separate clean-context reviewer tries to refute the work. No agent grades its own work. Fail closed.
  4. Every added agent earns its fan-out. Default to a single strong agent. More agents add intelligence (extra perspectives, verification), not parallel actions.

Plus the two non-negotiables: human-gate anything irreversible/MAJOR (push, deploy, send, real-money trade, mass-delete, new spend), and log an honest receipt of every kept/killed decision with the real number.

Fleet Mode receipt demo

The gate (run in order)

  1. Classify stakes. trivial: light check, ship. non-trivial: full gate. irreversible: full gate plus human approval first.
  2. Decide fan-out (bias-to-NO). Single strong agent by default; fan out only for read-heavy, parallelizable work that exceeds one context window.
  3. Write single-threaded.
  4. QC gate. Deterministic checks, then an independent reviewer that tries to refute. Escalate high-stakes output to a different-model judge.
  5. Human-gate MAJOR items.
  6. Log a receipt.

Full operational spec: SKILL.md.

Dependencies

The skill references the superpowers:* skill family (for dispatching parallel agents and requesting independent review). If superpowers skills are unavailable, substitute: fan-out decisions manually, run tests before any model review, get a colleague to review independently. The receipt script has zero dependencies (Python stdlib only).

Install (as a Claude Code skill)

git clone https://github.com/Jott2121/fleet-mode ~/.claude/skills/fleet-mode

Claude Code auto-discovers it. Invoke it (or let it auto-apply) on any non-trivial change. The receipt tool stands alone too:

python3 scripts/append_receipt.py \
  --task "add retry to uploader" --tried "exponential backoff" --verdict kept \
  --why "cut timeout errors to zero in a 200-run soak" --metric errors --value 0

Appends an append-only row to KILLLOG.md: the honest ledger of what you kept, killed, and why. See KILLLOG.md for a sample of the format.

Why it exists

Evaluate without prejudice; adopt only on measured proof. Receipts over hype.

The multi-agent discourse is full of impressive-sounding fan-out diagrams that, measured, lose to a single careful agent. Fleet Mode is the discipline that keeps the wins (parallel reading, independent review, hard gates) and drops the cargo-culting (parallel writers, self-grading, unbounded agent swarms). It scales to the task, not to the hype.

License

MIT, copyright 2026 Jeff Otterson. See LICENSE.

Ver o original no GitHub

Bom para

Relacionados