#MCP Server Integration

Model Context Protocol (MCP) lets your AI assistant call external tools -- query databases, search the web, operate files. This guide shows how to configure and use MCP in Aino LifeOS.

#Step 1: Understand Configuration Levels

Aino LifeOS supports three MCP configuration levels, from highest to lowest priority:

#Step 2: Add Your First MCP Server

#Option A: Configure in Your Vault (Recommended for Beginners)

Create a .mcp.json file in your vault root directory:

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/dir"]
    }
  }
}

Save and restart Aino LifeOS -- the server loads automatically.

#Option B: User-Level Configuration

To make a tool available across all vaults, edit ~/.claude/mcp.json:

{
  "mcpServers": {
    "web-search": {
      "command": "npx",
      "args": ["-y", "@anthropic/mcp-server-web-search"]
    }
  }
}

User-level and app-level MCP servers are not exposed to ordinary models by default. Turn on Global MCP servers in Settings → AI → MCP when you want Aino to start and show those servers.

#Option C: App-Level Configuration

Edit <userData>/mcp.json (good for team-wide configs):

{
  "mcpServers": {
    "company-tools": {
      "command": "node",
      "args": ["./tools/mcp-server.js"]
    }
  }
}

#Step 3: Verify Server Status

After configuration, open Aino LifeOS:

1. Servers are auto-discovered and loaded at startup
2. Check the MCP management panel -- status should show "Running"
3. Click a server name to see its tool list
4. In the AI chat tool panel, temporarily enable or disable a whole server or individual tools

If a server crashes, Aino LifeOS auto-restarts it.

#Step 4: Use MCP Tools in AI Chat

Once MCP servers are loaded, AI chat automatically gains these tool capabilities:

1. Open AI Smart Chat
2. Ask normally -- AI will automatically invoke MCP tools as needed
3. For example, with web-search configured, ask: "Search for the latest React 19 changes"

The AI calls the search tool, gets results, and replies with organized information.

#Write and Delete Permissions

MCP tools have extra permission controls so ordinary models do not mutate data unexpectedly:

• Settings → AI → Permissions → Allow write / update tools enables MCP tools whose names look like write, update, or create operations
• Settings → AI → Permissions → Allow delete tools enables MCP tools whose names look like delete/remove operations

These switches do not replace Claude Code / Codex CLI permission systems, but they do control which MCP tools are exposed to Aino's ordinary-model chat.

#Transport Protocols

#Stdio (Recommended)

Runs as a subprocess, ideal for local tools:

• Fast startup, low latency
• No network configuration needed
• Processes auto-managed by Aino LifeOS

#HTTP

Connect to remote MCP servers, ideal for team sharing:

• CORS support
• Customizable via middleware

#Built-in MCP HTTP Server

Aino LifeOS includes a built-in MCP HTTP server (default port 1479), allowing external tools to connect inbound. Uses include:

• Browser extension connecting to Aino LifeOS via HTTP
• Other apps integrating with your note data
• Custom automation workflows

#Common Configuration Examples

#Filesystem Access

{
  "mcpServers": {
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/you/Documents"]
    }
  }
}

#Web Search

{
  "mcpServers": {
    "web-search": {
      "command": "npx",
      "args": ["-y", "@anthropic/mcp-server-web-search"]
    }
  }
}

#Related

• AI Smart Chat -- MCP tools are automatically called by AI during chat