What problem does it solve? Operating Blue Yonder WMS (RedPrairie / JDA / Luminate) involves actions that move physical stock, overwrite on-hand inventory, or re-shape warehouse-wide behavior, and a wrong move causes real losses, phantom inventory, or mis-shipped goods. This Skill gives an AI agent the operator judgment to classify each warehouse action as read, reversible write, committing write, or destructive, and to pause consequential ones for human approval. ## Core Features & Use Cases - Risk-classified operation matrix: Maps every Blue Yonder WMS operation family (wave release, task confirmation, inventory adjustment, cycle count, ship confirm, policy and MOCA changes) to a gating level with named-approver rules for destructive actions. - Object and state model: Explains on-hand vs available, ILPN/OLPN nesting, allocation, wave vs waveless flow, and the RF-directed task lifecycle so the agent reasons about state rather than nouns. - Gotchas and recovery playbooks: Documents 24 causal failure chains (short picks, double-posted confirmations, stale on-hand, hold releases) plus recovery tables showing what can and cannot be undone. - Use Case: When asked to adjust inventory to match the ERP, the agent recognizes this as a destructive direct write to the book, requires a physical re-count and named approver, and refuses to force-match a host discrepancy. ## Quick Start Ask the agent how to safely release a wave or post an inventory adjustment in Blue Yonder WMS and it will classify the action, state the blast radius, and gate it appropriately.
