Household Voice Control
mikecutillo/household-voice-control
Quebre silos de voz: Alexa → IA Local → Qualquer Dispositivo
Ver no GitHubCamada de interface de voz que conecta Amazon Alexa (e futuramente outros assistentes) a um Sistema Operacional de IA local, permitindo que comandos de voz acionem qualquer serviço integrado. Usa AWS Lambda como ponte de nuvem, webhook local em Next.js para roteamento de intenções e Claude para interpretação de comandos em linguagem natural.
Instalação
npx convez-ai mikecutillo/household-voice-control1 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
- Controlar luzes Philips Hue e dispositivos de rede por voz sem estar preso ao ecossistema Amazon
- Enviar mensagens Discord ou escrever em Notion via comando de voz natural
- Usar Claude como fallback para interpretar comandos complexos não mapeados em slots rígidos
Veredito

Projeto bem arquitetado que resolve um problema real: aprisionamento de voz em silos de fabricantes. A combinação de Alexa → Lambda → webhook local + Claude como parser de linguagem natural é elegante e prática para automação doméstica soberana. Código privado limita auditoria, mas a documentação técnica é clara e o escopo é focado.
README
A tradução para o português deste README ainda está na fila.
Household Voice Control
A voice-interface layer that bridges Amazon Alexa (and, in progress, other assistants) to a local AI Operating System — so a household voice command can trigger anything in AIOS instead of being limited to the vendor's walled garden of skills.
This repo is a public overview. The running code is private.
What it is
Consumer voice assistants are siloed — Alexa talks to Alexa things, Siri talks to Apple things, Google talks to Google things. This project replaces that with a single passthrough layer: voice commands hit a custom skill, which forwards the intent to a local webhook running inside AIOS. From there, any connected service — lights, Discord posts, Notion writes, AI queries — becomes voice-addressable.
What it does
- Custom Alexa Skill that captures arbitrary spoken intents and forwards them to AIOS
- Webhook router inside AIOS that parses intent, resolves the target capability, and executes
- Natural-language fallback — if no explicit intent matches, the raw utterance is sent to Claude for interpretation and command generation
- Response synthesis — AIOS composes a spoken reply and hands it back to Alexa for TTS
- Device control bridge — ties into local Hue lights, router SOAP API, and curfew/device state
Architecture
graph LR
User[🗣️ Voice] --> Alexa[🎙️ Alexa]
Alexa --> Skill[☁️ Custom Skill]
Skill --> Lambda[λ AWS Lambda · TS]
Lambda --> Webhook[📥 AIOS webhook]
Webhook --> Intent{Intent match?}
Intent -->|yes| Cap[Capability handler]
Intent -->|no| Claude[🤖 Claude parse]
Claude --> Cap
Cap --> Devices[💡 Hue · 📡 Router · 💬 Discord · 📝 Notion]
Cap --> Reply[🔊 Spoken reply]
Reply --> Lambda
Software
| Layer | Tech |
|---|---|
| Voice frontend | Amazon Alexa Skills Kit, custom skill with open-ended slots |
| Cloud bridge | AWS Lambda (TypeScript) |
| Local bridge | AIOS webhook endpoint (Next.js API route) |
| Intent reasoning | Anthropic Claude for fallback parsing and response generation |
| Device layer | Philips Hue API, router SOAP, Discord webhooks |
What this demonstrates
- Vendor-neutral voice architecture — the voice stack is a transport, not a lock-in
- LLM-as-parser — leaning on Claude to handle the long tail of natural commands that rigid slot templates can't express
- Bidirectional flow — voice → local AI → connected device → spoken confirmation, all round-tripping cleanly
Stack
Related in the AIOS Portfolio
- BMO Discord Agent — Discord-native family AI companion; common destination for routed voice intents
- AIOS — The receiver; Next.js dashboard orchestrating 16+ household and business modules
- Spellbook Extension — Chrome MV3 extension turning page context + voice into Claude-executed action plans
Part of the AIOS portfolio. See the profile README for the full system map.