Telegram AI Agent Bot

Production-ready Telegram bot with intelligent agent features:

• Integrates with Telegram via aiogram
• Three MCP tools: Web Search, Weather, GitHub
• Context memory (SQLite)
• Vector knowledge base (Chroma + embeddings)
• Dockerized, with docker-compose and optional systemd service

Features

• /help: usage guide
• /health: liveness and readiness checks
• Session-scoped context memory with persistence
• Semantic search in vector knowledge base
• MCP-like tools exposed via mcp_client.py (Web Search, Weather, GitHub)

Requirements

• Python 3.9+
• Ubuntu 22.04+ / Debian 12+
• Docker 24+ and Docker Compose v2

Quick Start (Local)

1. Copy env file:

   cp .env.example .env

2. Fill .env with your credentials.
3. Create and activate venv, then install:

   python3 -m venv .venv && source .venv/bin/activate
   pip install -U pip
   pip install -r requirements.txt

4. Run bot:

   python -m src.bot

Docker Run

cp .env.example .env
docker compose up -d --build

Systemd Autostart

To enable automatic startup on system reboot:

# Copy service file to systemd directory
sudo cp mcp-telegram-bot.service /etc/systemd/system/

# Reload systemd
sudo systemctl daemon-reload

# Enable service to start on boot
sudo systemctl enable mcp-telegram-bot

# Start service
sudo systemctl start mcp-telegram-bot

# Check status
sudo systemctl status mcp-telegram-bot

# View logs
sudo journalctl -u mcp-telegram-bot -f

Repository Structure

README.md
docker-compose.yml
.env.example
requirements.txt
src/
  bot.py
  mcp_client.py
  memory_manager.py
  knowledge_base.py
conversations/
  cursor_chat_log.md
report.md

MCP Tools

• Web Search: via DuckDuckGo Lite HTTP API (no key) or SerpAPI (if provided)
• Weather: Open-Meteo (no key) or OpenWeather (if provided)
• GitHub: REST API with PAT (optional read-only)

Notes

• Chroma persists to ./data/chroma/
• SQLite DB persists to ./data/memory.sqlite