MCP tasks require a Pro plan subscription.
What is MCP?
The Model Context Protocol is an open standard that enables AI agents to interact with external tools and data sources. When you connect an MCP server to your agent, it can discover and use the tools that server provides, seamlessly incorporating them into conversations.Configuration
Server URL
The HTTPS endpoint of your MCP server. This is where your agent will connect to discover and invoke tools. Example:https://mcp.example.com/mcp
Transport Protocol
Choose how your agent communicates with the MCP server:- Streamable HTTP (Recommended) - Modern streaming protocol with better performance and reliability
- SSE (Server-Sent Events) - Alternative transport for servers that don’t support streamable HTTP
Custom Headers
Add optional HTTP headers that will be sent with every request to the MCP server. Useful for:- Authentication tokens
- API keys
- Custom routing information
Trust Acknowledgment
Before saving your MCP configuration, you must acknowledge that:- Your data may be shared with the third-party MCP server
- You trust the server and its operators
- You understand the security implications of this integration
Available Tools
Once connected, your agent automatically discovers the tools available on the MCP server. These tools appear in your configuration after a successful connection test, showing:- Tool name - The identifier used to invoke the tool
- Description - What the tool does (if provided by the server)
Testing Your Connection
Before saving, use the Test Connection button to verify:- Your agent can reach the MCP server
- The server responds correctly
- Available tools are discovered
- Not configured - No URL entered yet
- Connected - Server reachable and tools discovered
- Connected, no tools - Server reachable but no tools available
- Connection failed - Unable to reach or authenticate with the server

