PallasBot/Pallas-Bot

What it solves

Pallas-Bot is a feature-rich, learning-based chat bot designed for group environments. It solves the problem of creating a bot that feels like a natural participant in a community—capable of learning from group chat history to mimic local speech patterns, while also providing advanced AI-driven features like long-term memory, knowledge retrieval, and game-specific data lookups.

How it works

Built on the NoneBot2 framework, the bot uses a "learning-repeater" core that observes group chat logs to automatically pick up speech styles and common phrases. It can be configured to interact with LLMs for complex queries when explicitly mentioned, while maintaining a distinct personality constrained by the group's unique communication style. It supports a plugin ecosystem, high-concurrency message processing, and provides a WebUI for administrative control and monitoring.

Who it’s for

This project is for community managers, developers, or enthusiasts who want to deploy a highly customizable, personality-driven bot in group chats (such as QQ). It is particularly well-suited for fans of the game Arknights, as it includes built-in tools for querying game data, searching lore, and remote-controlling game assistants.

Highlights

• Learning-based interaction: Automatically learns from group chat logs to mimic speech styles and repeat messages.
• Federated corpus: Optional participation in a community-wide pool of chat snippets to enrich the bot's conversational variety.
• Game-specific integration: Built-in support for Arknights data retrieval and remote control of the MaaAssistantArknights (MAA) tool.
• Flexible AI capabilities: Supports LLM-based multi-turn conversations with long-term memory, web searching, and tool calling.
• Extensible architecture: Includes a plugin store, WebUI dashboard for management, and support for horizontal scaling via multiple workers.