Secretary MCP

Model Context Protocol Personal Assistant on Telegram

Category: ai-automation | Role: AI Automation Developer — Built 55-node workflow, configured MCP tool endpoints, connected Google Calendar and Gmail connectors, and implemented Telegram agent loop.

Project Overview

Secretary MCP is an autonomous personal productivity agent implemented in a 55-node n8n workflow (Secretary_MCP_Telegram.json). Operating over a Telegram Bot interface, the assistant leverages OpenAI GPT-4 and the Model Context Protocol (MCP) to expose modular external tools to the language model. The agent tracks dialogue context with Buffer Window Memory and can schedule or query Google Calendar events, search and draft Gmail messages, and perform live external research across SerpAPI, Wikipedia, and HackerNews to provide concise, factual answers.

Key Engineering Highlights

  • 55-Node Production Workflow: Structured workflow implementation in Secretary_MCP_Telegram.json
  • Model Context Protocol (MCP): Connects standardized tool definitions for autonomous agent execution
  • Google Workspace Integration: Automatically queries and creates Google Calendar events and handles Gmail
  • Multi-Source Web Research: Equips the agent with SerpAPI, Wikipedia, and HackerNews lookup tools
  • Contextual Memory: Maintains dialogue continuity across conversational turns via Buffer Window Memory

Technology Stack

n8nModel Context ProtocolOpenAI GPT-4Telegram BotBuffer Window MemoryGoogle CalendarGmailSerpAPIWikipediaHackerNews