Using Skills

Using Skills

AgentLogistics skills are written for AI agents that support progressive disclosure: load the smallest useful instruction first, then load references only when the task requires them.

Basic Use Pattern

1. Identify the logistics task.
2. Choose the matching atomic skill or skillset.
3. Read the package SKILL.md.
4. Load referenced files only when needed.
5. Gather required inputs and evidence.
6. Apply the skill procedure.
7. Produce the output contract.
8. State assumptions, missing evidence, calculations, and approval boundaries.

Example Skills

• calculate-reorder-point
• investigate-inventory-discrepancy
• plan-picking-wave
• audit-freight-charge
• map-wms-tms-integration
• perform-logistics-root-cause-analysis
• build-daily-warehouse-plan
• triage-temperature-excursion
• prepare-dangerous-goods-shipping-research
• plan-customs-broker-handoff

When To Use Skillsets

Use a skillset when the request is role-shaped or crosses multiple atomic skills. Examples:

• warehouse operator receive-to-ship workflow;
• inventory control discrepancy investigation;
• transportation coordinator freight analysis;
• logistics systems analyst integration review;
• distribution manager KPI and labor review.

When To Use Specializations

Use a specialization when the request depends on jurisdiction, product class, transport mode, customs lane, regulatory source, food safety, dangerous goods, or other non-universal constraints.

Output Expectations

Good outputs separate:

• facts;
• assumptions;
• missing inputs;
• source-backed requirements;
• calculations and units;
• source conflicts;
• operational recommendations;
• owner handoffs;
• approval boundaries.

Safety Boundary

A skill may prepare analysis, research, planning, and escalation packets. It must not approve legal, regulatory, engineering, food safety, dangerous-goods, customs, financial, customer, equipment, operator, or live-system decisions.