What is CoreLayer?

CoreLayer is a desktop AI command layer powered by Jarvis. It is not a generic chatbot, not an IDE copilot, and not a cloud automation tool. It sits locally between your personal apps, tools, models, and MCP servers so Jarvis can coordinate work through explicit permissions.

Key Capabilities

• Multi-model routing across MiMo, Groq, OpenRouter, Ollama, and OpenAI-compatible providers
• MCP-first integration with external tools and servers
• Unified tool calling from native modules, MCP, skills, and REST adapters
• Permission guard with risk-based execution and audit logs
• Voice pipeline with wake word, ASR, streaming TTS, and barge-in interruption
• Three storage modes (local SQLite, Supabase, PostgreSQL) hot-swappable at runtime
• Command palette (Alt+Space) for quick actions
• Task management with AI-powered natural language queries
• Reading list tracking with AI recommendations
• Daily/weekly review summaries with pattern recognition

Architecture at a Glance

Tauri 2 Desktop App
  └─ React 19 + Vite + Tailwind 4 Frontend
     └─ Tauri IPC Bridge
        └─ Node.js Daemon (Hono)
           ├─ Model Gateway (Vercel AI SDK)
           ├─ Tool Registry
           ├─ Permission Guard
           ├─ MCP Client Manager
           └─ SQLite / Supabase / PostgreSQL

CoreLayer uses a three-layer proxy model: the React frontend communicates with the Node.js daemon through Tauri IPC, and the daemon manages all AI orchestration, tool execution, and data storage.

Who is CoreLayer For?

• Technical users who want local control over AI workflows
• Indie developers building with MCP and local AI tools
• Power users experimenting with multi-model routing and voice interfaces
• Future contributors interested in the open-source desktop AI space

Next Steps

• Install CoreLayer — get the desktop app running
• Quickstart — connect your first model and MCP server
• Architecture — understand the system design