Operations Research Skill — Development skill for Claude Code

operations-research — an Agent Skill for OR & optimization

A [Claude Agent Skill](https://docs.claude.com) that helps Claude work like a trained operations-research practitioner across three modes:

• Solve (primary) — turn a real-world decision problem into a precise mathematical model, solve it in Python, and interpret the result (including sensitivity/duality).
• Explain — answer questions about OR theory and algorithms.
• Tutor — teach OR concepts step by step.

It covers linear and integer programming, network flows, scheduling/combinatorial models, nonlinear programming, and the stochastic side (queueing, simulation, inventory, Markov chains/MDPs, decision analysis), and solves with PuLP, Pyomo, SciPy, OR-Tools, or NetworkX — picking the right tool per problem and falling back gracefully when a solver isn't installed.

What's inside

operations-research/
├── SKILL.md                        # orchestrator: mode detection + the solve workflow
├── references/
│   ├── problem-taxonomy.md         # classify a problem → model class → tool → source
│   ├── formulation-patterns.md     # the modeling toolkit (logic, big-M, linearization)
│   ├── solver-guide.md             # PuLP/Pyomo/SciPy/OR-Tools/NetworkX code + fallback
│   ├── network-models.md           # transport/assignment/shortest-path/flow
│   ├── duality-sensitivity.md      # shadow prices, reduced costs, ranging
│   ├── beyond-lp.md                # IP, dynamic programming, NLP, game theory
│   ├── stochastic-models.md        # queueing, simulation, inventory, Markov/MDP, decisions
│   ├── worked-examples.md          # end-to-end solved problems (the output template)
│   └── book-map.md                 # topic → textbook → chapter navigation
└── scripts/
    └── pulp_report.py              # clean status/objective/duals report for PuLP solves

Install

This is a standard Agent Skill. You can either:

• distribute the packaged operations-research.skill file (it's a zip