← ClaudeAtlas

thinking-in-systemslisted

A systems-thinking skill based on "Thinking in Systems" (Donella Meadows). Draw stock-and-flow diagrams and feedback loops, diagnose "why fixing it makes it worse" (watching only flows, ignoring delays, treating symptoms, eroding goals), and use the 12 leverage points to locate the level of intervention. Treat workflows and projects as systems. Use when the user says "the change had no effect", "the metric won't move", "where's the leverage point", "is this a vicious cycle", "what do we do about a systemic problem" (at the level of organization / process / metrics); also for analyzing process or organizational dynamics. For focus / task overload use bandwidth-and-deep-work.
Trustbaselab/trustbase-skills · ★ 1 · AI & Automation · score 74
Install: claude install-skill Trustbaselab/trustbase-skills
# Thinking in Systems · An Operating System for Structure and Loops **In one sentence: the behavior of a system is produced by its structure (stocks, flows, feedback loops, delays, non-linearity), not by any single element or by anyone's intent** — so when "fixing it makes it worse", the first thing to change is our understanding of the structure, not the amount of force applied. Based on the ideas of *Thinking in Systems* (Donella H. Meadows, 2008). ## When to use - The user describes a **recurring** problem: the harder we try the worse it gets, the change had no effect, we push one thing down and another pops up - Analyzing the dynamics of a system / process / organization: why won't this metric move, why does it oscillate, why do the strong get stronger - Planning an intervention: change the numbers? the rules? the goals? the paradigm? - Diagnosing metric distortion: all the KPIs are green but the outcome is bad, or the standard has been quietly lowered - Multiple roles pulling against each other, resource allocation frozen in place, dependence on external firefighting - The agent needs a system-level view of a workflow, project structure, or team division of labor **Lightweight mode**: when the only question is "is this a reinforcing or a balancing loop" or "which leverage level is this", output only "a one-sentence conclusion + the loop verdict + the leverage level"; when the user asks "why does fixing it make it worse" but wants only a first hypothesis, output the