KiraAI, a modular, multi-platform AI virtual being that connects Large Language Models (LLMs), and various chat platforms (QQ, Telegram) with a virtual being centered architecture.
- Multi-adapter messaging: QQ, Telegram...
- Customizable LLM providers and models
- Sending multiple messages at once
- Virtual being centered design
- Function calling
- Persistent memory
- Centralized logging and prompt management
core/: orchestration for config, LLMs, prompts, memory, loggingdata/: runtime dataconfig/: INI/JSON configs for adapters, models, providers, stickers
scripts/: convenience launchers
Important
This project is in active development, and breaking changes may occur.
- Python 3.10+
- Windows, macOS, or Linux
- Platform credentials/tokens for enabled adapters (QQ, Telegram, etc.)
- Python dependencies in
requirements.txt(install viapip install -r requirements.txt)
- Clone this repository.
- Create and activate a venv(virtual environment).
- Install dependencies:
pip install -r requirements.txt. - Prepare configuration files under
data/config/
Example (cmd):
python -m venv .venv
.\.venv\Scripts\activate.bat
pip install -r requirements.txtExample (Bash/Linux):
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txtKey configuration files in data/config/:
providers.ini: LLM/TTS provider credentials and endpointsmodels.ini: model names, parameters, and defaultsadapters.ini: enable/disable platform adapters and their tokens
Check adapter settings for detailed documentation of messaging platforms
Modify the following if you want to get more custom experience:
bot.ini: core bot settings and runtime togglessticker.json: sticker mapping used by adapterstools/*.ini: per-tool configs liketavily.ini,ntfy.ini,bili.ini
Also, modify /data/persona.txt to make your unique virtual being!
You can start KiraAI via:
- CMD/PowerShell:
python main.py - Batch script:
scripts\run.bat - Linux script:
scripts/run.sh(make executable first)
Make Linux script executable and run:
chmod +x scripts/run.sh
scripts/run.shPlatform-specific entry points:
- Telegram adapter:
adapters/telegram/tg.py - QQ adapter:
adapters/qq/qq_reply.py
KiraAI/
core/ # Config/LLM/logging/memory/prompt managers
data/ # Memory store, stickers and configuration
scripts/ # Launch scripts
main.py # Main launcher
- Check logs in
log.log - Verify INI paths and sections match the enabled adapters and models.
- Ensure platform tokens are valid.







