ReAct Agent Consultation — Reason-Act-Observe Design Report

I will assess your agent use case and deliver a written ReAct agent design report covering whether the ReAct (Reason + Act) framework is the right pattern for your task, the optimal thought-action-observation cycle configuration, tool design for ReAct compatibility, stopping condition design, and implementation approach using LangChain or a custom ReAct loop. ReAct is the dominant agent reasoning pattern for tool-using agents because it produces auditable reasoning traces showing exactly why the agent chose each action — but it is not universally optimal; tasks requiring parallel tool execution benefit from different patterns, and tasks with deterministic decision trees are better served by structured workflows.

The report covers your task and tool requirements, ReAct suitability assessment, thought-action-observation cycle design, stopping condition recommendation, and implementation guidance.