MCP Servers
Connect external tools and services to your workflows using the Model Context Protocol.
MCP (Model Context Protocol) is a standard protocol that lets you connect external tools and services to your AI workflows. Think of it as a universal adapter for AI integrations.
What MCP Enables
- Connect to databases, APIs, and file systems
- Use pre-built integrations from the MCP Store
- Build custom tool servers for your specific needs
- Share tools across multiple workflows
How It Works
Agent → MCP Server → External Service (API, Database, etc.)
- Configure — Set up authentication and connection
- Connect — Add MCP server to your workspace
- Use — Agents call tools exposed by the server
MCP Gateway
The MCP Gateway in Lasius provides a centralized interface for managing MCP connections:
| Component | Purpose |
|---|---|
| Providers | Authentication mechanisms (API Key, OAuth, etc.) |
| MCP Store | Browse and connect pre-built servers |
| Connections | Active profiles and connection status |
| My Servers | Configured servers with available tools |
Transport Types
MCP supports multiple connection methods:
| Transport | Use Case |
|---|---|
| STDIO | Local command-line tools |
| SSE | Server-Sent Events for streaming |
| WebSocket | Real-time bidirectional communication |
| HTTP | Standard REST API calls |
Available Integrations
Pre-built MCP servers in the store include:
- Gsuite — Google Workspace (Drive, Sheets, Gmail)
- Hubspot — CRM and marketing automation
- Brave Search — Web search capabilities
- Fetch — HTTP requests to any URL
- Calculator — Mathematical operations
Custom MCP Servers
You can bring your own MCP servers for:
- Internal APIs
- Proprietary systems
- Custom business logic
- Private data sources
Using MCP in Workflows
Once configured, MCP servers appear as tools for your agents:
- Add an Agent node to your workflow
- In Tools, select your configured MCP server
- The agent can call any tool from that server
Setup Guide: Configure MCP Gateway