MCP Integration

Kaily.ai supports Model Context Protocol (MCP), an open standard designed to establish secure, two-way, real-time connections between AI-powered tools and backend systems.

Overview of Model Context Protocol (MCP) architecture with SSE and REST endpoints

What is MCP?

Model Context Protocol (MCP) enables AI agents to trigger actions in real-time by listening to external system events through a Server-Sent Events (SSE) stream.

There are two key components in the MCP architecture:

• MCP Servers: These expose actions or data endpoints via SSE and REST.

• MCP Clients: These are AI tools (like Kaily) that consume events from the server and act accordingly.

  info

  Kaily acts as an MCP client, allowing AI agent to execute tasks based on real-time data received from a configured MCP server.

---

How Kaily Uses MCP

When you connect an MCP server to AI Agent, it:

• Establishes a persistent SSE connection.
• Listens for events emitted by the server.
• Maps those events to defined actions inside AI Agent environment.
• Executes these actions securely based on user input or workflows.

---

Connecting an MCP Server

To integrate an MCP server in your AI Agent:

1. Navigate to the MCP section in the left sidebar to view your existing MCP server connections.

2. Click on the plus icon and select "Custom MCP" button to create a new MCP server connection.

   Select the MCP section from the left sidebar to manage your MCP server integrations

3. Fill in the configuration form with the required details:

   • Name: Enter a descriptive display name for your MCP server. This name will help you identify the server in your dashboard (e.g., "Zapier MCP", "Composio Server", "My Custom MCP").

   • SSE URL: Enter the Server-Sent Events (SSE) endpoint URL provided by your MCP server. This URL establishes the real-time event stream connection between AI Agent and your MCP server. Make sure to copy the complete URL from your MCP server provider.

   Click the Add MCP button to open the configuration form for a new MCP server

4. Review your configuration to ensure both the Name and SSE URL are correctly entered, then click Save to create the connection and establish communication with your MCP server.

Once saved, AI Agent will:

• Validate the server connection and test the SSE endpoint.

• Automatically fetch available Actions from the MCP server.

• Display them in the UI for further configuration or workflow use.

  View and add available AI actions retrieved from your MCP server to configure workflows

---

Supported MCP Servers

Here are some commonly supported MCP server platforms:

Provider	Description
Composio	Low-code platform offering ready-to-use action endpoints with SSE support.
Boltic	Data action and transformation server with real-time triggers.
Zapier	Event-based automation flows (supports webhook relay as SSE via proxy).
Pipedream	Developer-friendly integration hub that can act as an MCP server.
n8n	Open-source workflow automation tool with event-based triggers.
Make (Integromat)	Visual automation platform supporting webhook-based events.

---

Example: Automating Tasks with Zapier MCP Server

Here’s a step-by-step example using Zapier’s MCP server:

1. Go to https://mcp.zapier.com and create a new MCP server.

2. Once created, copy the SSE URL provided by Zapier.

3. Open Kaily platform and navigate to the MCP section.

4. Click "Add MCP" and paste the SSE URL in the configuration.

5. Give the server a name like "Zapier MCP" and click Save.

   info

   AI Agent will automatically fetch all supported actions from the Zapier server after you save.

6. Once the connection is established, AI Agent will display available actions like:

   • send_email
   • post_slack_message
   • create_google_calendar_event

7. You can now create workflows or configure AI Agent to respond to real-time events like:

   • Sending a welcome email when a form is submitted.
   • Posting a message to Slack when a task is created in your CRM.
   • Notifying users when a GitHub issue is labeled.

This setup empowers your AI Agent to act on external triggers without writing custom code, enabling seamless integration across tools.

---