OpenHands Repository Analysis Prompt

AI coding agent architecture review

• You need an agent to map how an AI coding system turns user tasks into actions, observations, and state transitions.
• You are evaluating workspace isolation, sandbox design, or human approval boundaries in a coding-agent repository.
• You want a reusable architecture review checklist for agent loops, runtime services, events, and integration surfaces.

• You only need a product overview or README summary.
• You are reviewing a small CLI without long-running agent state, event streams, or sandboxed execution.
• You want to publish production prompt rows directly without maintainer review.