CoPaw NEW

CoPaw — co personal agent workstation is an open-source AI assistant built on the Agentscope framework for deploying modular agents locally or in the cloud.It supports local LLM execution (llama.cpp, MLX for Apple silicon), self-hosted inference, and cloud APIs through a unified model layer.

The agent core separates prompts, hooks, tools and memory so developers can extend or replace components and assemble custom agents and skills via CLI.Multi-channel chat integration connects one assistant to platforms such as DingTalk, Feishu (Lark), QQ, Discord and iMessage, with support for custom channel plugins.

Persistent long-term memory and a heartbeat scheduler enable proactive tasks, automated workflows and cron-like skill execution for recurring jobs.Built-in skill engine, channel registry and message queue provide reliable message processing, discoverable skills and CLI commands for installing, configuring and managing integrations.

Typical use cases include personal productivity (email summarization, report generation), content workflows, research aggregation, desktop assistance and habit tracking while maintaining data control through local deployment options.

Key Features

• Agentscope-based modular agent deployment for local or cloud environments
• Unified model layer supporting local LLM execution (llama.cpp, MLX), self-hosted inference, and cloud APIs
• Modular agent core separating prompts, hooks, tools, and memory, extensible and composable via CLI
• Multi-channel chat integration with built-in connectors (DingTalk, Feishu, QQ, Discord, iMessage) and custom channel plugin support
• Persistent long-term memory and heartbeat scheduler for scheduled/cron-like skill execution and automated workflows

Use Cases

• Build a self-hosted AI assistant with copaw to run local LLMs securely, integrate with Slack, email and web chat via multi-channel support, and leverage long-term memory for personalized, privacy-preserving customer or employee support
• Automate recurring operations by composing modular agents and schedulers in copaw to run CLI tools, extract and aggregate reports on-premise, and persist outputs into long-term memory for auditability and reproducible workflows
• Create custom research and development assistants using copaw's plugin and model-layer architecture to orchestrate multi-step agents, perform sensitive data analysis locally, and deploy collaborative workflows in cloud or self-hosted environments

Who is it for?

• Software developers
• Ml engineers
• Devops engineers
• Open-source contributors
• Content creators