route-optimization

What it does

The skill is framed as an expert in transportation route optimization and vehicle routing problems. It aims to help design optimal delivery routes that minimize costs, reduce travel distance, improve service levels, and maximize fleet utilization.

How it works

It provides an Initial Assessment to gather Business Context, Operational Constraints, Network Characteristics, and Service Requirements. It then describes a Route Optimization Framework with Problem Classification covering TSP, VRP, VRPTW, CVRP, MDVRP, VRPPD, and Dynamic VRP. For Implementation, it details multiple Optimization Methods:

• Exact Methods: Branch and Bound; Mixed-Integer Programming (MIP) with a Python snippet using PuLP to model VRP and extract routes.
• Heuristic Methods: Nearest Neighbor, Clarke-Wright Savings Algorithm, and Sweep Algorithm, each with accompanying Python code to construct routes, compute total distance, and return route structures.
• Metaheuristic Methods: Simulated Annealing with code for generating neighbors, calculating total distance, and iterating with temperature cooling.

The skill emphasizes concrete steps, including choosing problem variants, constructing routes, and evaluating total distance. It explicitly names tools like "Claude Code" in the declared tools and shows code blocks with Python to implement VRP techniques.

When to use it

The skill should be used when asked to optimize delivery routes, solve vehicle routing problems, or minimize transportation costs, especially if mentions include route planning, delivery optimization, VRP, TSP, multi-stop routing, route sequencing, or dispatch optimization. It also mentions fleet sizing and management context, and last-mile-delivery references for related challenges.

What it can touch

The skill references the tool "claude-code" as a declared tool. The code examples demonstrate usage of Python libraries (e.g., PuLP) for optimization modeling and standard Python structures for route construction and distance calculations.

Caveats

It lists limitations of exact methods (e.g., Branch and Bound is feasible only for small problems <50 stops) and notes typical constraints like capacity, time windows, and depot structures within VRP variants. It also states that the Clarke-Wright, Nearest Neighbor, and Sweep methods are heuristics with trade-offs between speed and solution quality. No additional licensing or risk statements beyond these are provided in the given text.