AgentScope
ActiveDescription
Open-source multi-agent framework from Alibaba, enabling the construction of observable and interpretable agents with rich distributed capabilities.
Key Features
- Built-in ReAct Agent — Create agents with reasoning and tool-use capabilities in 5 minutes
- Tool & Skill System — Built-in Bash, Grep, file I/O tools with custom extension support
- Multi-Agent Orchestration — Flexible multi-agent collaboration via Message Hub
- MCP/A2A Native Support — Compatible with Model Context Protocol and Agent-to-Agent communication
- Agent-as-Service — FastAPI-based multi-tenant, multi-session agent service deployment
- Real-time Streaming — Event-based streaming output for real-time UI rendering
Use Cases
💡 Building personal AI assistants with file operations and code execution capabilities
💡 Deploying multi-agent collaboration systems for complex research or data analysis tasks
💡 Wrapping agents as API services for frontend applications and third-party systems to call
💡 Integrating external tools and data sources via MCP protocol to extend agent capabilities
💡 Building interactive agent platforms with Web UI for team usage
Strengths & Limitations
✅ Strengths
- • Actively maintained, recent updates
- • High community interest (30.5k stars)
- • Permissive open-source license (Apache-2.0)
- • Established track record (2 years in production)
Categories
Quick Start
pip install agentscope
from agentscope.agent import Agent
from agentscope.tool import Toolkit, Bash, Read, Write
from agentscope.model import DashScopeChatModel
from agentscope.message import UserMsg
import os, asyncio
agent = Agent(
name="Friday",
system_prompt="You are a helpful assistant.",
model=DashScopeChatModel(
api_key=os.environ["DASHSCOPE_API_KEY"],
model="qwen3.6-plus",
),
toolkit=Toolkit(tools=[Bash(), Read(), Write()]),
)
async def main():
async for evt in agent.reply_stream(UserMsg("Tony", "Hi!")):
print(evt)
asyncio.run(main())