Systems thinking
Structured thinking skills for AI agents: brainstorming, decision support, systems thinking, and strategy. Built for Claude Code & Cursor.
npx -y skills add ysskrishna/ai-agent-skills --skill systems-thinkingAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use this skill when the user asks for systems thinking (including naming it or directing use/apply/run with obvious misspellings; decisive) or wants to see how parts connect—mapping interdependencies, feedback loops, delays, stocks and flows, and leverage points before recommending action. Use when they worry about unintended consequences, holistic views, root causes beyond single blame, or how choices propagate across teams, products, or policies, even if they never say "feedback loop" or "stock and flow". Skip for single-step linear tasks, lone-variable calculations, or fixes that need no map of interactions or incentives.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.5 KB, as published. Nobody here has run it
Systems Thinking
See the whole before optimizing parts. End with leverage-aware recommendations.
How to run it with this skill: one clearly headed section per phase in this order: Boundary → Structure → Dynamics → Delays → Leverage → Synthesis. Do not skip Structure before Leverage.
Setup (run before starting)
In one short block:
- System in focus — what is inside the boundary?
- Default pass — Boundary → Structure → Dynamics → Delays → Leverage → Synthesis (state this line)
If the boundary is unclear, ask at most 3 scoping questions in one message, then proceed. Note any remaining gaps or working guesses in plain language (no bracket tags in Setup).
If the user already named a proposed intervention, you may spend less text on Boundary but still must not skip Structure before Leverage.
The Phases
Boundary
What is in vs out of the system for this analysis? State the purpose of the system from a stakeholder view (one sentence).
Structure
Elements (stocks: things that accumulate; actors; resources) and flows (rates in/out). Use short bullet pairs: From → To with what moves.
Dynamics
Identify at least one reinforcing (R) and one balancing (B) loop if plausible. Format:
Loop [R|B]: … — Mechanism: …
If a loop does not apply, say so in one line and justify.
Delays
Where is time lag between action and effect? How does delay change behavior (overshoot, oscillation, learned helplessness)?
Leverage
Leverage point: … — Why it matters: … — Risk of backfire: …
Prefer interventions that change rules, information flows, or goals over name-and-shame unless evidence supports it; favor levers that shift incentives rather than only blaming individuals.
Synthesis
- System story — one paragraph in plain language (no jargon wall)
- Non-obvious consequence — at least one
- Recommended moves — 2–3 actions compatible with the map (not contradicted by delays/loops)
Execution Rules
- Do not jump to solutions before Structure and at least a light Dynamics pass.
- Avoid linear blame ("X is stupid") as a stopping point; translate into structural incentives if you mention behavior.
- Diagrams optional; bullets must stand alone.
Checklist (verify before responding)
- Setup: system in focus + default pass stated
- Boundary explicit; purpose stated
- Structure uses elements + From → To flows
- Dynamics: R/B loops justified or explicitly N/A
- Delays considered where time matters
- Leverage tied to structure/dynamics, with backfire risk
- Synthesis tells a coherent system story