supply-chain-ops

Real inventory and supply-chain operations math for a Claude Code skill — standard textbook formulas, every one cited, every result auditable by hand.

A Claude Code skill for small business owners, ops managers, and anyone doing their own inventory planning without an enterprise ERP system.

Why this exists

EOQ, reorder points, ABC analysis, and demand forecasting are all well-established operations management formulas — this skill doesn't invent anything. It just makes them available at the terminal, with every number shown so you can check the math yourself. See reference/formulas.md for full derivations, sourcing, and the real limitations of each method.

What it does

Economic Order Quantity

python3 scripts/eoq.py --annual-demand 12000 --order-cost 75 --holding-cost 3.50

Real output:

=== Economic Order Quantity ===
  EOQ:                    717.1 units/order
  Orders per year:        16.7
  Days between orders:    21.8

=== Cost at EOQ (this is the minimum) ===
  Annual ordering cost:   $  1,254.99
  Annual holding cost:    $  1,254.99
  Total annual cost:      $  2,509.98

(Ordering cost equaling holding cost at the EOQ isn't a coincidence — it's the real minimum point of the classic Wilson EOQ cost curve, a built-in sanity check on the math.)

Reorder point & safety stock

python3 scripts/reorder_point.py --avg-daily-demand 40 --max-daily-demand 65 \
  --avg-lead-time-days 7 --max-lead-time-days 12

Uses actual observed demand/lead-time variability, not a made-up buffer percentage.

ABC inventory classification

python3 scripts/abc_analysis.py --csv items.csv

Standard Pareto-based classification from a CSV of item_id,annual_units,unit_cost — tells you which SKUs actually deserve tight control.

Demand forecasting

python3 scripts/demand_forecast.py --csv sales_history.csv --method both --periods-ahead 3

CSV needs a single demand column, oldest period first (extra columns are ignored). Moving average and exponential smoothing — no ML, no black box, both auditable by hand. Honest about their real limitation: neither captures trend, and the tool says so.

Install

git clone https://github.com/mikeblythe/supply-chain-ops ~/.claude/skills/supply-chain-ops

Claude Code picks it up automatically next session.

Honesty rules this skill follows

• Every formula cited to standard operations-management methodology, not invented
• Every result shows its work — nothing is a black box
• Forecasts with fewer than 3-4 periods of history are flagged as unreliable, not presented with false confidence

License

MIT