N8N-Test-Agents

August 28, 2025 ยท View on GitHub

A collection of AI agent system prompts designed to test various MCP (Model Context Protocol) tools with N8N workflows.

Agent Index

AgentMCP ToolFunction TestedDescription
Email SendingEmail MCPSend emailsTests email composition and sending functionality to Hannah about Netflix recommendations
Email RetrievalEmail MCPRetrieve/search emailsTests inbox search, filtering, and email organization capabilities
CalendarCalendar MCPEvent managementTests calendar event creation, scheduling, and meeting coordination
Google ContactsGoogle Contacts MCPContact managementTests adding, updating, and organizing contact information
Google DocsGoogle Docs MCPDocument operationsTests document creation, editing, formatting, and sharing
Google DriveGoogle Drive MCPFile managementTests file upload, organization, folder creation, and sharing
Notion DocsNotion MCPKnowledge managementTests Notion page creation, database management, and workspace organization
Task ManagementTask MCPProductivity workflowsTests task creation, status updates, project organization, and deadline tracking

Usage

Each agent folder contains a system-prompt.md file that defines:

  • The agent's role and personality
  • Specific task instructions for Daniel
  • Expected tool execution patterns
  • Output formatting templates
  • Confirmation requirements

Testing Structure

Email MCP Functions

  • Sending (/email-mcp/sending/) - Outbound email composition and delivery
  • Retrieval (/email-mcp/retrieval/) - Inbox search and email organization

Other MCP Tools

Each remaining tool focuses on its primary function set:

  • Calendar - Event scheduling and management
  • Contacts - Contact database operations
  • Docs - Document creation and editing
  • Drive - File storage and organization
  • Notion - Knowledge base management
  • Tasks - Productivity and project tracking

Integration with N8N

These system prompts are designed to work with N8N workflows that:

  1. Load the appropriate system prompt
  2. Connect to the corresponding MCP server
  3. Execute tool functions based on user input
  4. Provide structured feedback and confirmations