convez.ai
MCP ServerDesenvolvimento WebIntermediário
aipachakutiqwan

Simple Streamable MCP

aipachakutiqwan/simple-streamable-mcp

MCP client-server chatbot streamável com Claude para busca de papers

Ver no GitHub
InicialPluginScore · de 0 a 100

Servidor MCP leve e streamável powered por Claude para criar chatbots cliente-servidor. Implementação de referência para interagir com papers acadêmicos do arXiv, demonstrando padrões de comunicação MCP com suporte a streaming de respostas.

Instalação

$npx convez-ai aipachakutiqwan/simple-streamable-mcp

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
0
Acompanhando
0
Saúde
57%
Último commit
há 1 ano
Versão
—
Issues abertas
0
Contribuidores
2

Casos de uso

  • Criar bots de pesquisa acadêmica com streaming de respostas
  • Implementar padrão MCP para aplicações que requerem comunicação cliente-servidor
  • Extrair e processar informações de papers do arXiv
  • Aprender arquitetura MCP com exemplo funcional de referência

Veredito

Implementação educacional clara do padrão MCP com suporte a streaming. Adequado como referência para entender arquitetura cliente-servidor, mas explicitamente não pronto para produção — requer revisão, refatoração e testes antes de qualquer deploy real.

README

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

simple-streamable-mcp

Welcome to the lightweight Streamable MCP client-server chatbot powered by Anthropic Claude as the LLM backend.

⚡ Introduction

This repository provides a simple implementation of a streamable MCP client-server chatbot that uses Anthropic Claude as the LLM backend. The chatbot is designed to answer questions about research papers on arXiv or extract specific information from them.

This implementation is a slight modification based on the publicly available MCP: Build Rich-Context AI Apps with Anthropic course.

:heavy_exclamation_mark: This repository code is not intended for direct use in production. Do not use it as-is. It serves as a reference implementation. You should review the logic, refactor the code to fit your requirements, and write appropriate unit tests before considering it for any production deployment.

:rocket: Setup

🌱 Create an environment and install dependencies

Check out the Contributing guide to learn how to set up the environment, install dependencies, and get started contributing to this repository.

🌱 Setting up env variables

You can use .env file for set the following enviromental variables.

export ANTHROPIC_API_KEY=SET_YOUR_API_KEY_HERE
export SERVER_CONFIG_FILE=config/server_config.yaml
export LOCAL_CONFIG_FILE=config/local_config.yaml
export LOG_CONFIG_FILE=config/log_config.yaml
export RUN_LOCALLY=False
export PORT=8001
export ANTHROPIC_MODEL=claude-3-7-sonnet-20250219
export MAX_TOKENS_MODEL=2024

🌱 Antrophic Claude API

  • Sign up here and set ANTHROPIC_API_KEY in your environment.

✅ How to Use

:pushpin: Locally

This version initializes the chatbot and creates an MCP instance locally using the local configuration file.

export RUN_LOCALLY=True
python src/client/mcp_chatbot.py

:pushpin: Client/Server

This version starts the MPC server and initializes an MCP client chatbot to communicate with it. Follow the next two steps to activate the Client/Server option.

Launch a terminal session and activate the MCP server.

export RUN_LOCALLY=False
python src/server/mcp_server.py

Open a new terminal and start the MCP client.

export RUN_LOCALLY=False
python src/client/mcp_chatbot.py

:pushpin: MCP Inspector

Enables activation of the MCP inspector interface for interaction with the MCP server.

Execute the following command, setting the STDIO parameters to establish a connection with the MCP server.

mcp dev src/server/mcp_server.py
  • Transport type: STDIO
  • Command: uv
  • Arguments: run --with mcp mcp run src/server/mcp_server.py

🤖 Chatbot

The chatbot will appear as shown below—this is an example of its interface in the terminal.

Hello! I'm here to help you search for academic papers on arXiv or extract information about specific papers.
(Write 'quit' to exit.)

Use: @papers to see available papers topics.
Use: @<topic> to search papers in that topic.
Use: /prompts to list available prompts.
Use: /prompt <prompt name> <arg1=value1> to execute a prompt.

Send a message:
Ver o original no GitHub

Bom para

Relacionados