What problem does it solve? Operating Manhattan Active Omni (OMS) without deep platform knowledge leads to costly mistakes: overselling inventory by committing against stale availability, double-charging customers through blind payment retries, or releasing fraud holds that end in chargebacks. This Skill gives an AI agent the operator judgment to classify every order, payment, and sourcing action by risk before executing it. ## Core Features & Use Cases - Read/Write/Destructive Classification: Maps every OMS operation family (commit, allocate, release, settle, refund, appeasement, cancel) to a gating tier so consequential actions pause for human approval. - Availability & Sourcing Judgment: Explains Available-to-Commit vs raw on-hand, soft vs hard reservations, sourcing rules, node capacity, and safety stock so promises are made against real network supply. - Payment & Returns Lifecycle: Covers authorization, settlement, expiry and re-auth, refunds, multi-tender, exchanges, and appeasements, including gateway reconciliation rules that prevent double-charging. - Use Case: When asked to cancel a customer's order, the Skill guides the agent to check each line's settlement state first: void the authorization if nothing shipped (reversible), or process a refund plus physical return if settlement already occurred (destructive, requiring approval). ## Quick Start Ask the agent to commit a new order in Manhattan Active Omni and it will first re-read network availability, classify the commit as a reserving action, and request approval before proceeding.
