Overview

This node connects to an MCP (Model Context Protocol) server to discover and interact with AI Agent capabilities. It supports discovering available tools, resources, prompts, and resource templates that an AI Agent can use. The node can also execute specific tools, read resources, and retrieve prompt templates. It is useful for integrating AI Agent functionalities into workflows, enabling dynamic discovery and execution of AI-related capabilities.

Properties

Output

JSON

• tools - Array of discovered tools with their name, description, JSON schema, and metadata (for 'discoverCapabilities' operation).
• rawResult - Raw result object returned when tool execution output is not a standard object.
• * - Other JSON outputs vary depending on the operation, such as lists of resources, prompts, resource templates, or results from executing tools or reading resources.

Dependencies

• Requires credentials for MCP client connection, either an API key credential for command line transport or SSE client credentials for SSE transport.
• Depends on external MCP SDK packages for client communication and schema validation.

Troubleshooting

• Connection errors may occur if MCP server credentials are missing or incorrect; ensure proper credentials are configured for the selected connection type.
• Parsing errors can happen if tool parameters are not valid JSON objects; verify JSON formatting of tool parameters.
• Resource URI or parameters may be required for certain operations; missing or invalid parameters will cause errors.
• If the MCP client fails to connect, check network accessibility and command or SSE endpoint correctness.
• Errors during tool execution or resource reading are logged with detailed messages; review logs for specific failure reasons.