MaestroMCP
Provides comprehensive MIDI control capabilities including listing MIDI ports, creating virtual ports for DAW connections, sending note on/off messages, control change commands, and executing complex musical sequences with precise timing control.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@MaestroMCP play a C major chord on channel 1"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MaestroMCP
Servidor MCP para IA Musical - Bridge entre assistentes de IA e sintetizadores MIDI.
🎯 Descrição
MaestroMCP é um servidor M### 5. Validar Ferramenta Debug
No MCP Inspector:
Conecte-se ao servidor
Vá para a aba "Tools"
Encontre
debug_hello_worldExecute com ou sem parâmetros
Verifique o output JSON
Related MCP server: Cloudflare Playwright MCP
🎵 Ferramentas MIDI Disponíveis
midi_list_ports
Lista todas as portas MIDI disponíveis (entrada e saída).
midi_create_virtual_port
Cria porta virtual MIDI para conexão com DAWs como GarageBand.
midi_send_note
Envia mensagens MIDI Note On/Off com controle de velocity e canal.
midi_send_cc
Envia mensagens Control Change para volume, pan, effects, etc.
midi_sequence_commands
Executa sequências complexas de comandos MIDI com timing preciso.
debug_hello_world
Ferramenta de debug para validação do servidor.
🧪 Exemplos de Uso
Via Claude Desktop
Liste as portas MIDI disponíveisToque a nota C4 com velocity 100 no canal 1Crie uma sequência musical tocando C4, E4, G4 com 500ms de duração cadaVia MCP Inspectorxt Protocol (MCP) desenvolvido em TypeScript que estabelece a fundação técnica para um ecossistema de IA musical. Implementa 5 ferramentas MIDI fundamentais usando WebMidi.js para controle de sintetizadores e dispositivos musicais.
📋 Características
✅ Servidor MCP completo - Implementação oficial usando
@modelcontextprotocol/sdk✅ 5 Ferramentas MIDI - Lista portas, criação virtual, notas, CC, sequências
✅ WebMidi.js Integration - Suporte nativo para MIDI em Node.js
✅ Integração Claude Desktop - Configuração completa para uso no Claude
✅ MCP Inspector Support - Testado e validado via interface oficial
✅ TypeScript nativo - Código tipado e moderno
✅ Error handling robusto - Tratamento adequado de erros
✅ Arquitetura escalável - Preparada para futuras funcionalidades musicais
🛠️ Instalação
Pré-requisitos
Node.js >= 18.0.0
npm ou yarn
Instalação
# Clone/navegue até o diretório do projeto
cd MaestroMCP
# Instale as dependências
npm install
# Compile o projeto
npm run build🚀 Uso
1. Executar o Servidor
# Modo desenvolvimento (compila e executa)
npm run dev
# Modo produção
npm start2. Configurar no Claude Desktop
Para usar o MaestroMCP com Claude Desktop, adicione a seguinte configuração ao arquivo claude_desktop_config.json:
macOS:
Localização: ~/Library/Application\ Support/Claude/claude_desktop_config.json
{
"mcpServers": {
"maestro-mcp": {
"command": "node",
"args": [
"/Users/SEU_USUARIO/Desktop/Encantaria Suite/Maestro2.0/MaestroMCP/dist/server.js"
],
"env": {
"PATH": "/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin"
}
}
}
}Windows:
Localização: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"maestro-mcp": {
"command": "node",
"args": [
"C:\\Users\\SEU_USUARIO\\Desktop\\Encantaria Suite\\Maestro2.0\\MaestroMCP\\dist\\server.js"
]
}
}
}Linux:
Localização: ~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"maestro-mcp": {
"command": "node",
"args": [
"/home/SEU_USUARIO/Desktop/Encantaria Suite/Maestro2.0/MaestroMCP/dist/server.js"
]
}
}
}Importante:
Substitua
SEU_USUARIOpelo seu nome de usuário realAjuste o caminho conforme sua instalação
Certifique-se de que o projeto foi compilado (
npm run build)Reinicie o Claude Desktop após adicionar a configuração
3. Verificar Integração
Após configurar:
Compile o projeto:
npm run buildReinicie o Claude Desktop completamente
Teste no Claude Desktop:
Liste as ferramentas MIDI disponíveisVerifique se as ferramentas aparecem:
debug_hello_worldmidi_list_portsmidi_create_virtual_portmidi_send_notemidi_send_ccmidi_sequence_commands
4. Testar com MCP Inspector (Desenvolvimento)
# Lança servidor + MCP Inspector
npm run inspectorO MCP Inspector abrirá automaticamente no navegador. Caso não abra:
Acesse
http://localhost:6274Use o token de autenticação exibido no terminal
3. Validar Ferramenta Debug
No MCP Inspector:
Conecte-se ao servidor
Vá para a aba "Tools"
Encontre
debug_hello_worldExecute com ou sem parâmetros
Verifique o output JSON
🧪 Ferramenta Debug
debug_hello_world
Descrição: Ferramenta de debug que retorna mensagem personalizada + metadata do servidor.
Input Schema:
{
message?: string // Mensagem personalizada (opcional, default: "Hello World")
}Output:
{
"success": true,
"message": "Hello World",
"timestamp": "2025-01-XX...",
"server_status": "operational",
"debug_info": {
"server_name": "MaestroMCP",
"version": "0.1.0"
}
}Exemplos:
# Via MCP Inspector - sem parâmetros
# Retorna: message="Hello World"
# Via MCP Inspector - com parâmetro personalizado
# Input: {"message": "Test successful"}
# Retorna: message="Test successful"📁 Estrutura do Projeto
MaestroMCP/
├── package.json # Dependências e scripts
├── tsconfig.json # Configuração TypeScript
├── src/
│ ├── server.ts # Servidor MCP principal
│ ├── tools/
│ │ └── debug.ts # Ferramenta debug
│ └── types/
│ └── index.ts # Types do projeto
├── dist/ # Build compilado
└── README.md # Esta documentação🔧 Scripts Disponíveis
npm run build- Compila TypeScript → JavaScriptnpm start- Executa servidor compiladonpm run dev- Compila + executa em modo desenvolvimentonpm run watch- Compila em modo watchnpm run inspector- Lança servidor + MCP Inspectornpm test- Alias para executar servidor
🐛 Debug e Troubleshooting
Servidor não conecta
Verifique se compilou sem erros:
npm run buildTeste execução direta:
node dist/server.jsVerifique logs de erro (stderr não interfere no protocolo MCP)
MCP Inspector não funciona
Verifique se está rodando na porta correta:
Inspector UI:
http://localhost:6274Proxy Server:
http://localhost:6277
Use token de autenticação exibido no terminal
Limpe cache do navegador se necessário
Ferramenta não aparece
Verifique se servidor iniciou sem logs de erro
Recompile após mudanças:
npm run buildRestart do MCP Inspector
🎼 Próximos Passos
Esta é a base técnica. Futuras versões incluirão:
🎹 Integração MIDI real - WebMidi.js para sintetizadores
🎵 Geração musical - Algoritmos de composição automática
🎧 Análise de áudio - Integração com IaraMCP
🎛️ Controle semântico - Mapeamento linguagem → parâmetros musicais
🎸 Suporte multi-instrumentos - Roland S-1 e outros
🎚️ Performance em tempo real - Resposta musical contextual
📄 Licença
MIT
🤝 Contribuindo
Fork o projeto
Crie uma branch para sua feature
Commit suas mudanças
Abra um Pull Request
MaestroMCP v0.1.0 - Fundação para o futuro da IA musical. 🎵✨
Available Tools
11 toolsaudio_analyze_realtimeAnalyze Audio RealtimeB
Perform real-time audio analysis with pitch detection and spectral features
Inputs: analysis_type, duration_ms, pitch_detection, spectral_features.
audio_midi_feedbackAudio MIDI Feedback SystemC
Continuous audio monitoring and intelligent learning feedback for AI improvement
Inputs: analysis_depth, include_raw_data, monitoring_duration.
audio_start_captureStart Audio CaptureC
Initialize real-time audio capture for MIDI-audio correlation analysis
Inputs: bufferSize, latencyHint, sampleRate, source.
configure_musical_intelligence🧠 Configure Musical Intelligence & Superior Audio SystemC
Configure global musical intelligence settings and superior audio feedback system (sample rate, buffer size, analysis capabilities)
Inputs: audio_feedback_level, auto_sustain, default_articulation, enable_harmonic_analysis, +9 more.
debug_hello_worldDebug Hello WorldC
Ferramenta de debug que retorna mensagem personalizada
Inputs: message.
midi_list_portsList MIDI PortsC
Lists all available MIDI input and output ports
Inputs: refresh.
midi_play_and_listenMIDI Play and ListenC
Execute MIDI sequence while analyzing audio response - core of the self-listening AI system
Inputs: outputPort (required), sequence (required), audio_analysis, correlation_mode, +1 more.
midi_play_phraseMIDI Play PhraseB
🎼 Play musical phrase with natural articulation and expression - converts 'C4 E4 G4' to legato/staccato
Inputs: notes (required), channel, outputPort, rhythm, +2 more.
midi_send_cc_enhanced🎛️ Enhanced MIDI Control Change with Superior Audio ContextA
Send MIDI CC with musical context awareness and real-time analysis of CC effects on audio output. REPLACES basic midi_send_cc with enhanced capabilities.
Inputs: controller (required), value (required), channel, outputPort.
midi_send_note_enhanced🎼 Enhanced MIDI Send Note with Superior Audio HearingA
Send MIDI note with automatic musical expression, real-time spectral/temporal/pitch analysis, and superior audio feedback. REPLACES basic midi_send_note with enhanced capabilities.
Inputs: note (required), action, channel, duration, +2 more.
midi_sequence_commands_enhanced🎵 Enhanced MIDI Sequence with Superior Audio AnalysisA
Execute MIDI sequence with musical intelligence and comprehensive temporal/spectral audio analysis during execution. REPLACES basic midi_sequence_commands and midi_sequence_with_articulation with enhanced capabilities.
Inputs: commands (required), outputPort.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 11 tools
The tools have some clear distinctions, such as between audio analysis and MIDI operations, but there is overlap and potential confusion. For example, 'midi_play_and_listen' and 'midi_sequence_commands_enhanced' both involve playing MIDI sequences with audio analysis, making their boundaries unclear. Additionally, the 'enhanced' tools replace basic ones, which could lead to misselection if both were present, but since they are replacements, this is mitigated.
Most tools follow a consistent snake_case pattern with clear prefixes like 'audio_' or 'midi_', which aids readability. However, there are minor deviations: 'debug_hello_world' uses a different naming style (no prefix, mixed language in description), and some tool names include emojis or lengthy descriptions that break the pattern slightly. Overall, the naming is mostly predictable and well-structured.
With 11 tools, the count is well-scoped for a server focused on audio and MIDI processing with AI integration. Each tool appears to serve a distinct purpose within this domain, such as capture, analysis, configuration, and playback, without feeling excessive or insufficient. This number supports comprehensive functionality while remaining manageable.
The tool set covers key aspects of audio-MIDI interaction, including capture, analysis, playback, and configuration, with enhanced versions for core operations. A minor gap is the lack of explicit tools for stopping or managing ongoing processes (e.g., stop capture or reset), but agents can likely work around this using existing tools. The domain is well-represented with no major dead ends.
Related MCP Connectors
Browser DAW for AI assistants: build multitrack songs with MIDI, effects, stems and generation.
Score video, make music as an agent: 100+ AI instruments play your MIDI, audio, stems, Logic/Ableton
Autonomous music production for AI agents with MIDI generation, QC and provenance.
Veena is an AI DAW with a personal music producer built in. Make songs in your studio from chat.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceA MIDI composition system that enables AI assistants to create music through FluidSynth, with capabilities for playing notes, creating melodies, managing tracks, and exporting audio.1-
- AlicenseNot gradedqualityNot gradedmaintenanceEnables AI assistants to control a browser through a set of tools, allowing them to perform web automation tasks like navigation, typing, clicking, and taking screenshots.Apache 2.0
- AlicenseNot gradedqualityNot gradedmaintenanceEnables AI assistants to control a browser through a set of tools, allowing them to perform web automation tasks like navigation, typing, clicking, and taking screenshots.Apache 2.0
- AlicenseNot gradedqualityNot gradedmaintenanceA server that enables AI assistants to control a browser through tools, allowing them to perform web automation tasks like navigation, typing, clicking, and taking screenshots.Apache 2.0