Larch — Framework de Automação Claude Code
character-ai/larch
Orquestração multi-agente para design, revisão e implementação de código
Ver no GitHubFramework de orquestração para automação de workflows com Claude, implementando design multi-agente, revisão de código colaborativa e processos de adjudicação baseados em votação. Combina skills compostos, painéis de revisores especializados e rastreamento de execuções para projetos de engenharia de software complexos.
Instalação
npx convez-ai character-ai/larch1 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
Casos de uso
- Automação de processos de design e code review distribuídos entre múltiplos agentes
- Resolução de conflitos técnicos através de painéis de votação especializados
- Rastreamento e auditoria de decisões de arquitetura em repositórios Git
- Workflows complexos de engenharia com checkpoint e validação entre etapas
Veredito

Larch é uma ferramenta sophisticated para equipes que precisam orquestrar workflows de desenvolvimento com IA. A curva de aprendizado é alta, mas oferece automação genuína de processos colaborativos complexos — adequado para projetos em escala que justifiquem a complexidade da configuração.
README
A tradução para o português deste README ainda está na fila.
Larch
Larch is a Claude Code workflow automation framework that orchestrates multi-agent design, code review, and implementation through collaborative AI-driven processes.
Table of Contents
- Setup
- Installation and Setup — plugin install, local development, agent setup recipes (Claude / Codex / Cursor), clean-main entry contract for
/implementand/design, what the plugin provides, thescripts/relevant-checks.shconsumer contract, prerequisites
- Installation and Setup — plugin install, local development, agent setup recipes (Claude / Codex / Cursor), clean-main entry contract for
- Reference
- Features
- Skills
- Aliases
- Review Agents — the unified
code-reviewerarchetype - Run Logs — committed
larch-logs/batch files, manifest, and tracking-issue comments - Topology Projection — stable anchors for cross-doc topology counts
- Linting — linters, Makefile targets, halt-rate regression harness
- Issue-Anchored Plan — live wire format for the /design ↔ /implement plan handoff and clarification round-trip
- Architecture and workflow
- Workflow Lifecycle — how skills compose end-to-end
- Agent System — parallel subagent orchestration
- Collaborative Sketches — the diverge-then-converge design phase
- Voting Process — the voting panel protocol
- Point Competition — reviewer scoring system
Features
- Multi-agent design planning, reviews, and adjudication — the configured sketch topology diverges, the dialectic judge panel resolves contested decisions, and the validation panel reviews the final plan.
- Voting-based review resolution — The YES/NO panel protocol adjudicates plan and code review findings.
- Reviewer competition scoring — Reviewers earn points based on finding quality; a scoreboard tracks accepted, neutral, and rejected findings.
- Tracked runs —
/implementwrites full run artifacts to committedlarch-logs/files and keeps the tracking issue slim with marker-keyed summary comments.
Skills
See docs/skills.md for full details on each skill.
Aliases
Shortcut skills shipped with the plugin. Each alias forwards to an existing skill with preset flags.
| Alias | Equivalent |
|---|---|
/im | /implement --merge (same public flags as /implement; requires positional <issue-N>) |