# Systems Thinking

> Systems Thinking

- Skill: `sethdford/systems-thinking` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sethdford/systems-thinking`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sethdford/systems-thinking/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sethdford (https://skillmd.com/u/sethdford)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sethdford/systems-thinking

---

# Systems Thinking

Treat problems as interconnected stocks, flows, and delays—not isolated levers. Local optimization often creates global side effects; look for structure before blaming people.

## When to Use
- Incentives, reliability, org dynamics, infrastructure, or policy where fixes bounce back or new issues appear downstream.

## Workflow
1. Map system boundaries: what is in-scope, inputs, outputs, and external pressures.
2. Identify key stocks (accumulations) and flows (rates in/out).
3. Trace feedback loops: label reinforcing (R) vs. balancing (B); note sign and delay.
4. Use Donella Meadows’ leverage hierarchy: constants and goals often beat parameter tweaks; mind unintended effects when moving high-leverage points.
5. Predict second-order and timing effects (delays can cause oscillation or overshoot).
6. Propose interventions at the strongest justified leverage point, with monitoring for side effects.

## Examples
**Example 1:** “Hire faster” increases throughput but raises onboarding load—map balancing loop on quality and a delayed R loop on attrition.

**Example 2:** Cost-cutting a vendor reduces spend (B) but can add incident rate (R) via slower patches—quantify delay before deciding.