QwenPaw

QwenPaw is an open source personal AI assistant for people who want an agent on their own computer or server, with persistent memory and access through chat apps. It runs on macOS, Windows and Linux, including Docker deployments, under the Apache 2.0 license.

Its built-in llama.cpp runtime runs QwenPaw-Flash models locally without an API key or cloud dependency. Ollama and LM Studio are also supported. With local inference, data stays on your machine. Cloud model providers such as OpenAI, Anthropic and Google Gemini require API keys and process model requests remotely; AgentScope Platform and ModelScope Studio also offer cloud deployments.

Memory combines the current working context with verbatim conversation history and a ReMe knowledge base. It turns conversations and resources into linked Markdown files you can read, edit and search. The browser console and terminal interface share the same agents, memory and sessions, with access through Telegram, Discord, DingTalk, Lark, WeChat, iMessage and QQ.

The assistant can schedule recurring tasks, process PDF and Office documents, gather information online, and read, edit, review and test project code. Skills, plugins and MCP connections extend its tools. Independent agents can work in parallel with their own memory and skills.

Execution controls include operating-system sandboxes, checks on tool calls, protection for sensitive files and scanning of skills before activation. Access policies can allow an action, deny it or require human approval.