agent-self-correction

Protocols for the agent to autonomously detect failures, analyze root causes, and attempt recovery strategies before requesting user intervention.