Build AgentScope Workflows with ReAct Agents and Tools

In this tutorial, we build a complete AgentScope workflow from the ground up and run everything in Colab. We start by wiring OpenAI through AgentScope and validating a basic model call to understand how messages and responses are handled. From there, we define custom tool functions, register them in a toolkit, and inspect the auto-generated schemas to see how tools are exposed to the agent.

We then move into a ReAct-based agent that dynamically decides when to call tools, followed by a multi-agent debate setup using MsgHub to simulate structured interaction between agents. Finally, we enforce structured outputs with Pydantic and execute a concurrent multi-agent pipeline in which multiple specialists analyze a problem in parallel, and a synthesizer combines their insights.

We install all required dependencies and patch the event loop to ensure asynchronous code runs smoothly in Colab. We securely capture the OpenAI API key and configure the model through a helper function for reuse. We then run a basic model call to verify the setup and inspect the response and token usage.

We define custom tool functions for mathematical evaluation and datetime retrieval using controlled execution. We register these tools into a toolkit and inspect their auto-generated JSON schemas to understand how AgentScope exposes them. We then simulate a direct tool call to validate that the tool execution pipeline works correctly.

We construct a ReAct agent that reasons about when to use tools and dynamically executes them. We pass user queries and observe how the agent combines reasoning with tool usage to produce answers. We also reset memory between queries to ensure independent and clean interactions.