convez.ai
MCP ServerDevOps & InfraestruturaAvançado
stigenai

MCP Servers - Orquestração de Servidores Model Context Protocol

stigenai/mcp-servers

Orquestrador de servidores MCP com Docker, Python/Node e automação CI/CD

Ver no GitHub
InicialPluginScore · de 0 a 100

Repositório que padroniza a containerização e orquestração de servidores MCP (Model Context Protocol) em Docker, suportando implementações em Python, Node.js e Docker customizado. Inclui servidores prontos como time, playwright e zettelkasten, com CI/CD automático via GitHub Actions e registro centralizado de metadados.

Instalação

$npx convez-ai stigenai/mcp-servers

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
0
Forks
0
Acompanhando
0
Saúde
50%
Último commit
há 1 dia
Versão
—
Issues abertas
2
Contribuidores
6

Casos de uso

  • Implantar e gerenciar múltiplos servidores MCP em ambiente containerizado
  • Integrar automação com Playwright para web scraping e testes de navegador
  • Centralizar conhecimento com Zettelkasten via MCP

Veredito

Solução madura e bem-arquitetada para quem precisa orquestrar múltiplos servidores MCP em produção. Requer familiaridade com Docker, CI/CD e shell scripting; ideal para equipes que integram MCP na infraestrutura existente.

README

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

MCP Servers

This repository contains Docker images for various MCP (Model Context Protocol) servers that can be deployed on the Stigen.io orchestrator platform.

Overview

MCP servers come in different varieties:

  • Python-based: Run via uvx (uv execution)
  • Node-based: Run via npx (node execution)
  • Docker-based: Pre-built containers

This repository provides a standardized way to containerize Python and Node-based MCP servers, while also supporting custom Docker images.

Repository Structure

servers/
├── python/           # Python-based MCP servers
│   ├── base/        # Base Python Docker image
│   └── time/        # Time server implementation
├── node/            # Node-based MCP servers
│   ├── base/        # Base Node Docker image
│   └── playwright/  # Playwright automation server
└── registry.json    # Server registry metadata

Supported Servers

Built-in Servers

ServerTypeDescriptionImage
timePYTHONProvides time and timezone functionalityghcr.io/stigenai/mcp-time:latest
playwrightNODEBrowser automation and web scraping with Playwrightghcr.io/stigenai/mcp-playwright:latest
zettelkastenPYTHONImplements the Zettelkasten methodology with SQLite storage for knowledge managementghcr.io/stigenai/mcp-zettelkasten:latest

Custom Servers

Users can also bring their own Docker containers by specifying serverType: "custom" and providing their own containerImage.

Building Locally

Prerequisites

  • Docker
  • Bash
  • jq
  • yq

Build a specific server

./scripts/build.sh time

Build all servers

./scripts/build-all.sh

GitHub Actions

The repository includes GitHub Actions workflows for build, validation, sync, and maintenance:

  • build-servers.yml: Detects changed MCP servers, validates registry/config integrity, and builds changed server images (PR + push + manual)
  • build-base-images.yml: Builds shared Python/Node base images (scheduled + manual + reusable)
  • build-registry.yml: Builds and publishes the registry OCI image when registry metadata changes
  • sync-mcp-repos.yml: Periodically syncs upstream MCP commit/release metadata into servers/registry.json and ignores archived/disabled upstream GitHub repositories (backfilling stale commit/release metadata to null)
  • cleanup-images.yml: Cleans up old container image tags in GHCR
  • pr-notify.yml: Sends Slack notifications for new ready-for-review PRs

Server Configuration

Each server has a config.yml file that defines:

  • name: Server identifier
  • type: python, node, or docker
  • version: Semantic version
  • environment: Environment variables with defaults
  • resources: CPU and memory requirements

Example:

name: mcp-time
type: python
version: 1.0.0
description: "MCP server that provides time and timezone functionality"
environment:
  - name: MCP_PORT
    default: "3000"
    required: false
    secret: false
resources:
  cpu: 0.25
  memory: 256

Environment Variables and Secrets

Servers can define environment variables in their config:

  • required: Whether the variable must be provided
  • secret: Whether the variable contains sensitive data
  • default: Default value if not provided

Secrets are stored encrypted per user/organization and injected at runtime.

Integration with Orchestrator

The server registry is available in two ways:

1. As an OCI Image (Recommended)

# Pull the registry image
docker pull ghcr.io/stigenai/mcp-servers-registry:latest

# Extract registry.json using the provided script
./scripts/get-registry.sh

# Or manually extract
docker create --name temp-registry ghcr.io/stigenai/mcp-servers-registry:latest
docker cp temp-registry:/registry.json ./registry.json
docker rm temp-registry

2. Via GitHub Raw URL

https://raw.githubusercontent.com/stigenai/mcp-servers/main/servers/registry.json

This allows the orchestrator to:

  • List available server types
  • Get default configurations
  • Validate server deployments
  • Auto-populate container images

Contributing

See CONTRIBUTING.md for dependency PR triage policy and contribution rules.

To add a new MCP server:

  1. Create a new directory under servers/[type]/[name]
  2. Add a Dockerfile
  3. Add a config.yml
  4. Update servers/registry.json
  5. Add a changelog entry in CHANGELOG.md for user-visible changes
  6. Test locally with ./scripts/build.sh [name]
  7. Submit a pull request

For release-note conventions, see docs/release/entry-process.md.

Changelog

License

[Add license information]

Ver o original no GitHub

Bom para

Relacionados