Mokei
TairuFramework/mokei
Framework MCP com suporte multi-provider, agentes e orquestração de contextos
Ver no GitHubToolkit TypeScript completo para criar, integrar e monitorar clientes e servidores usando Model Context Protocol (MCP). Suporta múltiplos provedores (OpenAI, Anthropic, Ollama, GGUF), gerenciamento de sessões, loops de agentes autônomos e observabilidade em tempo real de tráfego MCP.
Instalação
npx convez-ai TairuFramework/mokei1 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
- Construir aplicações com agentes autônomos que executam ferramentas via MCP
- Integrar múltiplos servidores MCP em uma única aplicação
- Monitorar e debugar comunicação MCP em tempo real
- Criar assistentes com acesso a múltiplos provedores de LLM (OpenAI, Claude, Ollama)
Veredito

Mokei é uma solução madura e bem-estruturada para trabalhar com MCP em TypeScript, com excelente suporte a múltiplos provedores e recursos avançados como decision flows e approval systems. Recomendado para desenvolvedores que precisam construir sistemas de agentes robustos com MCP.
README
A tradução para o português deste README ainda está na fila.
Mokei
TypeScript toolkit for creating, interacting with, and monitoring clients and servers using the Model Context Protocol (MCP).
Features
- MCP Server & Client - Full protocol implementation
- Protocol Revisions - Serves and speaks both
2025-11-25and2026-07-28, selected per context or negotiated automatically - Multi-Context Host - Manage multiple MCP servers with tool namespacing
- Session Management - High-level chat abstraction with tool calling
- Agent Loop - Automatic tool execution with configurable approval
- Model Providers - OpenAI, Anthropic, Ollama, and local GGUF (llama.cpp) integration
- Monitoring - Real-time observation of MCP traffic
- CLI - Interactive chat and server inspection tools
Quick Start
pnpm add @mokei/session @mokei/session-node @mokei/openai-provider
import { NodeSession } from '@mokei/session-node'
import { OpenAIProvider } from '@mokei/openai-provider'
const session = new NodeSession({
providers: {
openai: OpenAIProvider.fromConfig({
apiKey: process.env.OPENAI_API_KEY
})
}
})
// Add an MCP server
await session.addContext({
key: 'myserver',
command: 'node',
args: ['my-mcp-server.js']
})
// Chat with tool access
const response = await session.chat({
provider: 'openai',
model: 'gpt-4',
messages: [{ source: 'client', role: 'user', text: 'Hello!' }]
})
Documentation
- Full Documentation - Complete guides and API reference
- Quick Start Guide - Get running in minutes
- docs/ - Local documentation
Packages
| Package | Description |
|---|---|
@mokei/context-server | MCP server implementation |
@mokei/context-client | MCP client implementation |
@mokei/host | Multi-context orchestrator |
@mokei/session | Portable chat and agent sessions |
@mokei/session-node | Node stdio session support |
@mokei/openai-provider | OpenAI integration |
@mokei/anthropic-provider | Anthropic Claude integration |
@mokei/ollama-provider | Ollama integration |
@mokei/llama-provider | Local GGUF inference via node-llama-cpp |
@mokei/http-client | MCP Streamable HTTP client transport |
@mokei/http-server | MCP Streamable HTTP server transport |
@mokei/flow-host | Mokei flow host: run decision flows in the background with approval, inbox and recovery |
@mokei/flow-host-node | Mokei Node flow host: durable storage and observability for decision flows |
@mokei/flow-client | Mokei flow client: portable flow control interface, types and errors for the flow daemon |
@mokei/host-desktop | Desktop dialogs, notifications and input inbox for Mokei hosts |
@mokei/decision-flow | Mokei decision flow node kind for flow graphs |
mokei | CLI tool |
CLI
mokei monitor # Start the monitor UI for MCP server traffic
mokei inspect # Inspect an MCP server (prints how it describes itself)
mokei chat -p openai # Interactive chat
mokei proxy <command> # Proxy an MCP server through the daemon
mokei daemon status # Manage the per-user daemon
mokei flows list # Inspect configured flows
mokei runs list # Inspect flow runs
mokei inbox list # Inspect pending inputs and approvals
Contributing
See AGENTS.md for development setup and code style guidelines.
License
MIT