Table of Contents

• About
• Features
• Requirements
• Installation
• Usage
• Project Structure
• Configuration
• Commands
• Development

About

GeminiCommBot is a full-featured Telegram bot for working with Google Gemini and OpenAI AI models. It allows users to leverage powerful language models and image generation directly in Telegram, with full support for personal configuration and API key management.

Features

• Google Gemini support: text generation, image analysis, image generation
• OpenAI support: integration with OpenAI API for text generation
• Key management: multiple API keys with automatic rotation
• Flexible configuration: change model, system prompt, generation temperature, and other parameters
• Memory system: Mem0 integration for long-term user memory
• Security: encrypted data storage in the database
• Admin panel: database management, log viewing, notification broadcasting
• Inline mode: use the bot via @GeminiCommBot in any chat
• Beta mode: support for testing new features

Requirements

• Python 3.10+
• Telegram Bot API token from @BotFather
• API keys:
  • Google Gemini API key (get one at aistudio.google.com/api-keys)
  • OpenAI API key (optional, for OpenAI features)
  • Mem0 API token (for memory system)

Installation

1. Clone the repository

git clone https://github.com/IgorVasilekIV/gemini_comm
cd gemini_comm

2. Create a virtual environment

python3 -m venv .venv
source .venv/bin/activate  # Linux/macOS
# .venv\Scripts\activate   # Windows

3. Install dependencies

pip install -r requirements.txt

Usage

Configure the .env file

Create a .env file in the project root (or copy from .env.example):

# === Bot Tokens ===
MAIN_BOT_TOKEN=xxxxxxxxxx:XXXXX...
TEST_BOT_TOKEN=xxxxxxxxxx:XXXXX...

# === Admin ===
ADMIN_ID=xxxxxxxxxx

# === Database ===
DB_ENCRYPTION_KEY=xxxxxx...

# === AI Services ===
MEM0_TOKEN=m0-xxxxx...

# === Trial (optional) ===
TRIAL_KEY=sk-xxxx
TRIAL_BASE_URL=https://api.xxx
TRIAL_MODEL_CLASSIFICATION=
TRIAL_MODEL_TEXT=

# === Sync Keys (optional) ===
SYNC_KEYS_URL=

# === File Upload (optional) ===
FILE_UPLOAD_URL=https://x0.at

# === Test Keys (optional) ===
TEST_GEMINI_API_KEY=

Generate database encryption key

python3 -c "from cryptography.fernet import Fernet; print(Fernet.generate_key().decode())"

Run the bot

Normal mode:

python3 main.py

Beta mode (with debug logs and test token):

python3 main.py --beta

Using Make (alternative):

make install    # install dependencies
make run        # normal mode
make run-beta   # beta mode
make dev        # install with dev dependencies
make lint       # run linter
make format     # format code
make test       # run tests

Project Structure

gemini_comm/
├── main.py                 # Entry point, bot initialization
├── config.py               # Application configuration
├── database.py             # SQLite database (with encryption)
├── utils.py                # Utilities (file uploads, formatting, memory)
├── requirements.txt        # Project dependencies
├── pyproject.toml          # Project metadata
│
├── handlers/
│   └── main_handler.py     # Main command handlers
│
├── ai/
│   ├── gemini.py           # Google Gemini API integration
│   └── openai.py           # OpenAI API integration
│
├── tests/
│   ├── test_gemini.py
│   └── test_openai.py
│
├── assets/
│   └── avatar.jpg
│
├── Dockerfile              # Docker build
├── docker-compose.yml      # Docker Compose configuration
├── Makefile                # Development commands
├── LICENSE                 # MIT License
├── README.md               # This file
└── README_RU.md            # Russian documentation

Configuration

User configuration system

Each user can configure parameters for Gemini and OpenAI via commands:

Gemini parameters:

• api_key — one or more API keys (comma-separated)
• model — model for text generation
• model_img — model for image generation
• system_instruction — system instruction for the model
• temperature — generation temperature (0.0-2.0)
• negative_prompt_img — negative prompt for image generation

Database encryption

All sensitive data (API keys) is stored in SQLite with Fernet encryption. The encryption key is loaded from the DB_ENCRYPTION_KEY environment variable.

Commands

Note

< > — required parameter | [ ] — optional parameter

User commands

Command	Description	Example
/start	Start the bot, greeting	/start
/cfg	View/edit Gemini configuration	/cfg
/openai	View/edit OpenAI configuration	/openai
/fcfg <provider> <parameter> <value>	Set a parameter	/fcfg gemini model gemini-2.0-flash
/news	View latest announcements from admin	/news

Admin commands

Command	Description	Example
/logs	Send the latest log file	/logs
/getdb <user_id>	View user configuration	/getdb 123456789
/deldb <user_id>	Delete user configuration	/deldb 123456789
/news [title ~ items]	View latest announcement or create a new one	/news New update ~ Added new model

Inline mode

Note

Use @GeminiCommBot in any Telegram chat

Command	Description
@GeminiCommBot cfg	Show Gemini configuration
@GeminiCommBot openai	Show OpenAI configuration
@GeminiCommBot fcfg	Set a parameter
@GeminiCommBot [Models]	View available Gemini models
@GeminiCommBot <request>	Send a request to Gemini

Development

Architecture

Command handling:

• Main router is in handlers/main_handler.py
• Uses aiogram 3.x with FSM (Finite State Machine) for state management
• Inline query support via InlineQuery

AI integration:

• ai/gemini.py — Google Gemini API with key rotation
• ai/openai.py — OpenAI API integration
• Automatic rotation between multiple API keys when quota is exceeded

Database:

• SQLite with Fernet encryption
• Stores each user's configuration and current API key index
• Functions in database.py for CRUD operations

Running tests

python3 -m pytest tests/ -v
python3 -m pytest tests/ --cov=.

Logging

Logging is configured via loguru:

• Normal mode: INFO level
• Beta mode (--beta): DEBUG level
• Logs are output to stderr

License

MIT License

Support & Contact

• Telegram: @gemini_comm
• Bug reports: create an issue in the repository
• Discussion: @GeminiComm