Quickstart Guide

This guide walks you through the minimum setup to get CoreLayer working with an AI model and an MCP server.

Step 1: Launch CoreLayer

Open the CoreLayer desktop app. The Control Center will appear with the initial setup wizard.

Step 2: Configure a Model Provider

Navigate to Settings → Models and add a provider:

Provider: Groq (recommended for fast local-like experience)
API Key: your-groq-api-key
Model: llama-3.3-70b-versatile

You can also use:

• MiMo — Xiaomi's model via OpenRouter
• OpenRouter — access to multiple providers
• Ollama — fully local models
• OpenAI-compatible — any provider with an OpenAI-compatible API

Step 3: Connect an MCP Server

Navigate to Settings → MCP and add a server:

{
  "name": "filesystem",
  "command": "npx",
  "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/your/workspace"]
}

This gives Jarvis access to read and write files in your workspace through the MCP protocol.

Step 4: Test It

Open the command palette with Alt+Space and type:

List the files in my workspace

Jarvis will:

1. Interpret your intent
2. Route the request through the Model Gateway
3. Resolve the filesystem tool from the Tool Registry
4. Execute the MCP tool call
5. Return the results

If the tool requires write access, the Permission Guard will pause and ask for approval.

What Just Happened?

Your request
  → Jarvis interprets intent
  → Model Gateway selects Groq/llama-3.3
  → Tool Registry resolves filesystem MCP tool
  → Permission Guard checks risk level
  → MCP Client executes tool call
  → Result returned to you

Next Steps

• First MCP Server — deeper guide to MCP integration
• Model Providers — configure multiple providers
• Permission Guard — understand the safety model