Supply Chain & Fulfillment AI

Capacity Planner

Capacity Planner is an ecommerce AI skill for Alireza Rezvani, built for teams working with Codex, Claude Code, OpenClaw. Use it to you need to prioritize stocking,…

Provider
Alireza Rezvani
Platforms
Codex · Claude Code · OpenClaw
View original link · GitHub

What this skill helps you do

Sizing tool for ops teams handling queued work — Support, CX, BizOps, IT ops, Finance ops. Built on Erlang-C queueing theory, Little's Law, and operational-leadership canon. Produces capacity sizing at 70/80/90% utilization against P50/P90/P99 demand, per-member utilization health metrics, and a 12-month quarterly hiring plan accounting for ramp curves, attrition, and manager-trigger thresholds. Deterministic, stdlib-only Python tools — no LLM calls in computation.

Install and get started

Copy the full instructions into your AI tool. Test one low-risk example before connecting real store data.

Original Skill instructions

You are an ops leader sizing queued-work teams. Intake P50/P90/P99 daily ticket volume. Run capacity_modeler.py with demand, AHT, SLA target, FTE, and shrinkage per --profile. Read 80%-utilization row as sizing point. Run utilization_analyzer.py to flag sustained >85% utilization (throughput-collapse risk) and spread >30pp (UNBALANCED). Run hiring_sequencer.py for 12-month plan with ramp, attrition, and manager triggers at 7 ICs/manager. Walk the forcing-question library one at a time. Never plan to 100% utilization, treat ramp as instant, or size to P50 only. If you only have averages, stop and pull the distribution — single-point demand estimates are the most expensive anti-pattern in ops.

Useful tasks

  • Annual ops capacity planning with Erlang-C sizing against P50/P90/P99 demand distributions
  • Quarterly re-sizing when demand shifts >15% or attrition spikes
  • Pre-budget headcount defense with CFO-ready Erlang-C math
  • Diagnostic when an ops team misses SLA — sizing, process, or bottleneck problem?
  • M&A or new-segment launch workforce modeling

How to use it

  • Always use at least 90 days of historical demand data for reliable P50/P90/P99 computation
  • Never plan to 100% utilization — Reinertsen Principle 12: throughput collapses well before
  • Read the 80%-utilization row as your sizing point, not the 100% one
  • Model priority tiers (P1/P2/P3) as separate queues with class-specific SLAs then sum
  • Front-load Q1-Q2 hiring (35%) — ramp curves make late-year hires ineffective

More skills for this workflow

Content checked: