Agentic Development · Reliability
A pass where an agent reviews its own previous output or action result and decides whether to revise it, retry it, or move on.
Anti-example: It's not the same as just re-running the same prompt again — reflection specifically judges the prior output against the goal before deciding what happens next.