ReAct agent

Overview

The prebuilt ReAct agent implements a classic pattern:

• Reason/Plan with an LLM planner step
• Act by executing a tool requested by the LLM
• Repeat until you have a final answer

In Coevolved, react_agent(...) is built on top of agent_loop(...), so it supports checkpointing, budgets, and interrupts.

Required state shape

ReAct expects state to be either a dict or a Pydantic model with fields that behave like:

• messages: list of chat messages (recommended)
• llm_response: the last LLMResponse from the planner
• tool_name, tool_args, tool_call_id: metadata for the current tool call
• tool_result: output of the tool
• final: final assistant text (used as the stop signal)

Planner (LLM step)

The planner must be a step created by llm_step(...) and should attach an LLMResponse into state (default key: llm_response).

When tool specs are present in LLMConfig.tools, the provider can return tool calls in LLMResponse.tool_calls.

Tools (tool steps)

Tools must be steps created via tool_step(...) (annotation kind="tool").

At runtime, the agent:

• Validates that all tools are Steps and are marked as tools
• Resolves tool calls by tool name
• Executes exactly one tool call per iteration (first call in the list)

Loop behavior and termination

ReAct stops when final is not None.

If the LLM requests a tool that does not exist, ReAct raises a KeyError.

If a tool runs but does not set tool_result (or your configured result key), ReAct raises an error so you can fix the tool wrapper.

Next steps

Building tools

Define schemas and wrap tools so ReAct can call them.

prebuilt.react reference

Full parameter list and default state keys.