CoPaw: The Complete Guide to Your Personal AI Assistant
What if you had a personal AI assistant that works across DingTalk, Discord, QQ, Feishu, iMessage — and more — all from a single self-hosted deployment? That's CoPaw. With 9,100+ GitHub stars and growing fast since its February 2026 launch, CoPaw is an open-source personal AI assistant built by the AgentScope team at Alibaba. Easy to install, deploy locally or in the cloud, with extensible skills and persistent memory.
What Is CoPaw?
CoPaw (Co Personal Agent Workstation) is an open-source personal AI assistant designed to be your digital companion across all your communication channels. It's built on the AgentScope framework and follows a simple philosophy: works for you, grows with you.
- Language: Python
- License: Apache 2.0
- Stars: 9,100+ ⭐
- Forks: 1,000+
- Latest Version: v0.0.5 (March 2026)
Core Capabilities
🌐 Every Channel
Connect your assistant to the messaging apps you already use:
- DingTalk — Enterprise messaging
- Feishu (Lark) — Workplace collaboration
- QQ — Social messaging
- Discord — Community chat
- iMessage — Apple ecosystem
- Twilio — Voice channel (new in v0.0.5)
- And more — extensible channel architecture
One assistant, multiple channels. Messages come in, responses go out, regardless of the platform.
🔒 Under Your Control
- Deploy locally or in the cloud (Docker, Alibaba Cloud ECS, ModelScope)
- Memory system — Persistent, personalized memory (powered by ReMe)
- Scheduled reminders — Cron-based heartbeat to any channel
- Your data stays yours — No external dependencies required
🛠️ Skills System
Custom skills auto-loaded from your workspace — no vendor lock-in:
- Social: Daily digest of hot posts (Xiaohongshu, Zhihu, Reddit), Bilibili/YouTube summaries
- Productivity: Newsletter digests to DingTalk/Feishu/QQ, contacts from email/calendar
- Creative: Describe your goal, run overnight, get a draft next day
- Research: Track tech/AI news, build a personal knowledge base
- Desktop: Organize files, read/summarize docs, request files in chat
- Explore: Combine skills and cron into your own agentic app
🤖 Local LLM Support
Run completely offline with no API keys:
- llama.cpp — GGUF models
- MLX — Apple Silicon optimized
- Ollama — Easiest local model management
- Cloud providers — DashScope, ModelScope, and more
Download and manage models directly from the Console UI:
copaw models download Qwen/Qwen3-4B-GGUF
copaw models # select the downloaded model
copaw app # start the server
Quick Start
pip install (Recommended)
pip install copaw
copaw init --defaults
copaw app
Then open http://127.0.0.1:8088/ for the Console.
One-Line Install (Linux/macOS)
curl -fsSL https://raw.githubusercontent.com/agentscope-ai/CoPaw/main/install.sh | bash
Docker
docker run -p 8088:8088 agentscope/copaw
ModelScope (Cloud)
One-click deployment on ModelScope.
Console UI
CoPaw comes with a web-based Console accessible at http://127.0.0.1:8088/ after startup. From the Console you can:
- Chat with your assistant
- Configure models and API keys
- Manage channels (add DingTalk, Feishu, QQ, Discord, etc.)
- Install and configure skills
- Manage memory and personalization
- Download and switch local models
MCP Support
CoPaw supports the Model Context Protocol (MCP), enabling integration with MCP-compatible tools and services. This makes it easy to extend CoPaw's capabilities with third-party tools.
v0.0.5 Highlights (March 2026)
- Daemon mode — Run as a background service
- Twilio voice channel — Voice interactions
- DeepSeek Reasoner support — Enhanced reasoning
- Agent interruption API — Interrupt long-running tasks
- Memory system upgrade — Improved persistence
- Windows one-click installation — Cross-platform support
- Docker configuration persistence — Persistent Docker deployments
CoPaw vs Alternatives
Category: This tool is a self-hosted personal AI assistant with multi-channel chat integration.
| Feature | CoPaw | Open WebUI | AnythingLLM |
|---|---|---|---|
| Focus | Multi-channel personal AI assistant | Web UI for LLMs | All-in-one AI platform |
| Stars | 9.1K ⭐ | 126K ⭐ | 55.8K ⭐ |
| License | Apache 2.0 | Custom | MIT |
| Chat Channels | ✅ DingTalk, Feishu, QQ, Discord, iMessage, Twilio, more | ❌ Web UI only | ❌ Web UI only |
| Local LLMs | ✅ llama.cpp, MLX, Ollama | ✅ Ollama, OpenAI API | ✅ Ollama, LMStudio, LocalAI |
| Skills System | ✅ Custom skills, auto-loaded, cron-based | Plugins | ✅ Custom agents |
| Memory | ✅ Persistent memory (ReMe) | Conversation history | Document memory (RAG) |
| Cron/Heartbeat | ✅ Built-in scheduled tasks | ❌ | ❌ |
| MCP Support | ✅ | ✅ | ✅ |
| Console UI | ✅ Web-based | ✅ ChatGPT-like UI | ✅ Desktop app + web |
| RAG | Via skills | ✅ Built-in | ✅ Core feature |
| Multi-User | Via channels | ✅ Full RBAC | ✅ Multi-user |
| Docker | ✅ | ✅ | ✅ |
| Voice | ✅ Twilio | ❌ | ❌ |
| Language | Python | Python | JavaScript |
| Team | AgentScope (Alibaba) | Community | Mintplex Labs |
When to choose CoPaw: You want a personal AI assistant that connects to your existing messaging apps (DingTalk, Feishu, QQ, Discord, iMessage) with scheduled tasks, skills, and persistent memory. The multi-channel architecture is unique — no other tool connects to this many Asian messaging platforms.
When to choose Open WebUI: You want the most polished chat interface for interacting with local LLMs. Massive community (126K stars), ChatGPT-like UI, built-in RAG, fine-grained user management. Best for teams who want a private ChatGPT replacement.
When to choose AnythingLLM: You want an all-in-one AI productivity platform with document chat (RAG), custom agents, desktop app, and no-code setup. Strong focus on document processing and knowledge management.
Why CoPaw?
The name says it all: Co (collaborative) + Paw (a partner always by your side). CoPaw isn't just another chatbot — it's designed to be a warm companion for your digital life. The multi-channel architecture means you can reach your assistant wherever you already communicate, and the skills + cron system lets you build autonomous workflows that run on your schedule.
Conclusion
CoPaw fills a unique niche in the self-hosted AI assistant space: it's the only open-source project that seamlessly bridges multiple messaging platforms (DingTalk, Feishu, QQ, Discord, iMessage, Twilio) with a powerful skills system, persistent memory, and local LLM support. Built by the AgentScope team at Alibaba, it's production-ready with Docker support, cloud deployment options, and a growing ecosystem of skills.