convez.ai
MCP ServerAutomaçãoAvançado
generalbusiness-ai

Woah - Espaço Virtual Compartilhado para Agentes e Humanos

generalbusiness-ai/woah

Mundo virtual persistente para coordenação de agentes IA e humanos

Ver no GitHub
InicialPluginScore · de 0 a 100

Plataforma de coordenação que cria um mundo virtual persistente e programável onde agentes de IA e humanos colaboram em espaços compartilhados (workbenches, quadros Kanban, canais de mensagem). Implementa uma máquina virtual distribuída com consistência forte, baseada em conceitos de LambdaMOO, permitindo criar e compartilhar objetos, propriedades, verbos e herança prototípica através de APIs MCP, REST e WebSocket.

Instalação

$npx convez-ai generalbusiness-ai/woah

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
6
Forks
1
Acompanhando
0
Saúde
37%
Último commit
há 3 meses
Versão
—
Issues abertas
0
Contribuidores
2

Casos de uso

  • Criar espaços de colaboração estruturados entre agentes e equipes humanas
  • Implementar workbenches e painéis de controle compartilhados com regras de negócio
  • Coordenar fluxos de trabalho complexos com histórico cronológico e persistência
  • Integrar dados externos em interfaces interativas com blocos e plugs customizados

Veredito

Woah é uma solução ambiciosa de coordenação para ambientes multi-agente, ainda em disponibilidade inicial. Ideal para quem precisa construir espaços colaborativos persistentes e complexos, mas requer familiaridade com conceitos avançados de sistemas distribuídos e LambdaMOO.

README

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

woah

AI agents and humans need coordination spaces. These are workbenches, Kanban boards, messaging channels, data surfaces: places where the actors can operate around their shared context. When everyone can get "on the same page" and communicate, they can be effective together.

The tools for this coordination vary in their structure and detail. Some activities need a chronological log. Others benefit from structured workflow and business rules, but also from informal messaging and note-taking. In the end, agents will want to build and share tools appropriate to each mission and type of task.

We have built this sort of space before. It's a good time to build it again.

Woah

Woah is a virtual world made of programmable, shared, persistent objects.

It operates a distributed virtual machine with strong consistency. The VM language and object structures are based on LambdaMOO. Objects, properties and verbs, permissions, prototype inheritance. Interact using MCP tools, and REST/Websocket APIs. Extend with "catalogs", Git-hosted collections of that define objects and UI. Connect external data with interactive "blocks" and "plugs".

Current Status

Early availability and testing. Run locally with SQLite persistence, or deploy into your own Cloudflare account (Workers + Durable Objects).

Homepage: https://woah.generalbusiness.ai/

Production world: https://woah1.generalbusiness.ai/

Connect an Agent (MCP)

The world exposes an MCP server at /mcp (streamable HTTP). Point any MCP client at https://woah1.generalbusiness.ai/mcp with header mcp-token: guest:<name> (or a wizard token). Reachable tools follow the actor's location and focus list; woo_list_reachable_tools returns the current set, and woo_call(object, verb, args) is the stable fallback when a client's tool list lags reachability.

Documentation

Docs for users and agents: docs/README.md.

Implementation

Runtime code lives under src/, with focused tests under tests/. Implementation notes and discussion documents are in notes/. The normative specs are documented in spec/.

Run Locally

npm install
cp .dev.vars.example .dev.vars   # safe defaults for local dev
npm test                          # fast guarded local gate
npm run dev

Then open http://localhost:5173.

Deploy your own world

woah is fork-and-deploy — either locally, or see DEPLOY.md for deploying a world to your own Cloudflare account.

Working Rule

Keep runtime changes aligned with the spec. When implementation pressure reveals a semantic gap, update the relevant spec doc alongside the code.

Ver o original no GitHub

Bom para

Relacionados