MaiBot

// Overview

MaiBot — MaiSaka — is an LLM-based interactive agent built on a different premise from most assistants: it is not trying to be a helpful task-completing bot. The project's stated aim is a conversational presence that feels warm and authentic rather than perfect or efficient, so replies are short or long chat turns instead of long markdown-heavy answers, and the bot is designed to know when to join a conversation and when to stay quiet.

It is built for group chats specifically. In a multi-person conversation MaiSaka picks up how the people around it speak, learns new slang and in-group vocabulary as it goes, and — drawing on personality theory — accumulates a longer-term picture of each person's information, likes, dislikes and behaviour. A plugin system with APIs and an event system is the documented extension path, and the project ships a WebUI with a plugin marketplace for installing adapters and extensions.

Deployment is self-hosted and deliberately low-friction: a one-click package for Windows sets up Python and runs a setup wizard without touching a command line, while Linux and macOS users can use the repository's Docker Compose stack, which runs the core service, a NapCat container for QQ, and a SQLite browser over the shared data volume. You need a computer with an internet connection (Windows 10+, Linux or macOS), at least 2 GB of free memory, and an LLM API key — the model provider is configured in the WebUI, not baked in. It requires Python 3.12+ and is GPL-3.0 licensed.

// What it does

• Conversational style tuned for group chat — variable-length casual turns rather than markdown-formatted answers
• Decides when to speak and when to stay silent instead of replying to every message
• Picks up the speaking style, slang and in-group vocabulary of the people in a conversation
• Long-running per-person understanding, informed by personality theory, that accumulates across conversations
• WebUI for model configuration, plugins and a plugin marketplace
• Plugin system with APIs and an event system for extensions
• Platform adapters — QQ via NapCat is the documented common case, with other platforms supported through adapters
• Self-hosted deployment: a Windows one-click package, or a Docker Compose stack with core, NapCat and a SQLite browser

// Getting started

You need Windows 10+, Linux or macOS, at least 2 GB of free memory, and an LLM API key. Windows users can skip the command line entirely with the one-click launcher.

git clone https://github.com/Mai-with-u/MaiBot.git
cd MaiBot
docker compose up -d

http://localhost:8001      # one-click / source install
http://localhost:18001     # docker compose (mapped port)

WebUI → model configuration → provider address + API key → select model

NapCat → enable forward WebSocket
WebUI → plugin marketplace → install + enable "NapCat adapter"

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

// When to use it

• Reach for it when you want a persistent presence in a group chat rather than a question-answering bot
• Reach for it to run a companion bot entirely on your own machine against your own API key
• Reach for it when the bot should hold context about the people in the room across weeks, not just the current thread
• Reach for it as a plugin-extensible base for chat-platform automations built on an event system

// How MaiBot compares

MaiBot alongside other open-source assistants & chatbots tools AI/TLDR tracks, ranked by GitHub stars.