What problem does it solve? Operating Manhattan Active Transportation Management involves actions with very different risk levels—tendering a load commits carrier capacity and spend, a parcel manifest close starts billing, and Freight Audit and Payment moves money out—yet they look similar in the UI. This Skill gives an AI agent the judgment to classify each operation and pause consequential ones for human approval before committing freight spend or paying invoices. ## Core Features & Use Cases - Operation gating matrix: Classifies every Active TM operation family (orders, loads, rating, tendering, booking, manifesting, FAP settlement) as read, reversible write, committing write, or destructive, with matching approval gates. - Edge-state and recovery playbooks: Covers waterfall auto-roll, spot bid timers, partial acceptance, continuous moves, duplicate loads, EDI 204/990 races, and how to recover from cancellations, payment reversals, and closed periods. - Use Case: Before tendering a load, the agent re-reads tender status and carrier eligibility, shows carrier + mode + lane + total cost, and routes the tender through a human approval gate instead of committing freight spend autonomously. ## Quick Start Ask the agent to tender a load on Manhattan Active TM and have it classify the action, re-read carrier eligibility, and pause for approval before sending the EDI 204.
