AgentScope Update in 2025-11-04

[DavdGao]

Dear Community,

Thank you for your continued support since the release of AgentScope 1.0. We remain committed to enhancing and improving the AgentScope project. Now we have some updates to share.

🎉 New Agent Implementations

We're excited to open-source two new yet powerful agent applications built on AgentScope:

• 🤖 Alias-Agent: A versatile LLM-empowered agent application featuring four specialized modes (Simple ReAct, Planner-Executor, Deep Research, and Browser-Use) to flexibly handle diverse real-world tasks within a secure sandbox environment.

  • https://github.com/agentscope-ai/agentscope-samples/tree/main/alias

• 📊 Data-Juicer Agent: An intelligent multi-agent system that enables natural language-driven data processing by seamlessly integrating AgentScope's agent orchestration capabilities with Data-Juicer's 200+ professional data operators.

  • https://github.com/agentscope-ai/agentscope-samples/tree/main/data_juicer_agent

🚀 Core Capabilities Expansion

• 🏋️ Agentic RL Support: Fine-tune agentic workflows built with AgentScope using Trinity-RFT with minimal code changes, while providing advanced features for experienced users.

  • https://github.com/agentscope-ai/agentscope/tree/main/examples/training/react_agent

• 💾 Enhanced Long-term Memory: AgentScope has integrated ReMe, a smart memory toolkit to provide personal, task, and tool-level long-term memory management.

  • https://github.com/agentscope-ai/agentscope/tree/main/examples/functionality/long_term_memory/reme

📦 Introducing AgentScope-Samples

A curated collection of ready-to-use agent implementations and full-stack applications built with AgentScope, showcasing practical use cases across diverse domains.

🔗 Repository: https://github.com/agentscope-ai/agentscope-samples

💡 We warmly welcome contributions from the community—share your innovative agents and be part of the ecosystem!

⚙️ Powerful Runtime Upgrades

We've made it easier to deploy and interact with agents:

• 📦 App-like Agent Deployment: Consistent behavior from local dev to production, supporting Docker, Kubernetes, and Alibaba Cloud Function Compute (FC).
• 🐍 Python SDK: Programmatic interaction with deployed agents.
• 🖥️ GUI & Desktop-enabled Sandboxes: VNC-powered graphical control plus desktop_url support in filesystem and browser sandboxes for seamless desktop access via your browser.
  • https://github.com/agentscope-ai/agentscope-runtime

We look forward to your feedback and continued engagement with the community.

Best regards,
The AgentScope Team