What problem does it solve? Operating Oracle Transportation Management (OTM) through an AI assistant is risky because actions like tendering a shipment, securing resources, or approving a freight invoice commit real money and carrier capacity, and automation agents can fire them with no human in the loop. This Skill gives the model the operator judgment to classify every OTM action and pause consequential ones for human approval. ## Core Features & Use Cases - Operation classification: Maps OTM operation families (order release, bulk plan, tender, booking, freight settlement, automation agents) into read / write / committing / destructive classes with gating rules. - Edge-state and recovery playbooks: Covers partial tender acceptance, continuous move tours, spot bid auto-awards, auto-pay/self-billing, domain partitioning, and recovery patterns for un-tendering, voucher reversal, and double-planned freight. - Use Case: Before letting an agent tender a shipment, the Skill forces a rate/plan review showing the buy cost, itinerary, and service provider, re-reads tender status and carrier eligibility at execute time, and requires a named approver for cancellations or voucher reversals. ## Quick Start Ask the assistant to tender an Oracle OTM shipment to a carrier and have it classify the action and confirm the buy cost, service provider, and lane before committing.