🧠 s8eagv1 – Modular Cognitive Agent Framework

s8eagv1 is a modular AI agent framework designed to run multi-tool LLM agents using MCP (Model Control Protocol). It supports dynamic tool composition, Google Sheets integration, Telegram-based agent control, and local embeddings.

Demo

---

🚀 Features

• 🧩 MultiMCP: Integrates multiple MCP tool servers over both stdio and http
• 🤖 Custom Agent: Structured agent planning and execution using AgentLoop
• 📄 Google Sheets Tools: Create, update, and share spreadsheets via an HTTP (SSE) MCP server
• 📡 Telegram Bot Interface: Send query to your agent via chat, simply send a message
• 🔍 Semantic Search: Embed and index local files using nomic-ai/nomic-embed-text-v1
• 🧠 LLM Interop: Use models like gemini-2.0-flash through your local orchestration
• 🔧 Extensible Tool Servers: Easily add new tools via Python MCP wrappers

---

🧪 Step 2: Set Environment

TELEGRAM_BOT_TOKEN=your_token_here
DRIVE_FOLDER_ID=your_google_drive_folder_id
GOOGLE_CREDENTIALS=service_account.json

set evironment variable to run sse MCP server for Google Sheet operations, refer this repo.

export DRIVE_FOLDER_ID=<folder_id_in_which_sheets_are_created>

Run Http (SSE) Google Spreadsheet MCP Server

cd s8eagv1/mcp_servers/mcp_google_sheets
python server.py

🚀 Step 3: Run the Telegram Bot

python TelegramBot.py