Diagnoses last night’s batch failures and proposes a re-queue plan. It never executes anything — every step below is a proposal for a human to accept, modify, or reject.
Nothing has happened yet. Start a run to see the agent work.
Findings and the remediation plan will appear here once the run produces its final answer.