System dynamics analysis
Analyze persistent or surprising system behavior caused by feedback loops, delays, accumulations, nonlinear effects, constraints, incentives, and adaptation. Use for recurring oscillation, backlog, capacity, adoption, policy resistance, second-order effects, or fixes that make a problem return or worsen. Produce a Dynamic System Model. Do not use for an isolated implementation defect, a static dependency map, or estimating one identified causal effect from data.From its SKILL.md
npx -y skills add SylphxAI/skills --skill system-dynamics-analysisAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 1 stars1 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.
SKILL.md
3.3 KB, 561 tokens by cl100k_base, as published. Nobody here has run it
System Dynamics Analysis
Explain behavior over time rather than decorating a static architecture with arrows. Read references/system-dynamics-methods.md when constructing loops, stock-flow models, simulations, or interventions.
Workflow
- Define the focal behavior, decision, time horizon, system boundary, and the evidence showing how the behavior changes over time.
- Draw a behavior-over-time view for the important variables. Distinguish an observed pattern from a hypothesized mechanism.
- Identify accumulations and their rates of change: queues, inventory, installed base, knowledge, debt, trust, capacity, cash, or other relevant stocks and flows.
- Connect reinforcing and balancing feedback loops. Mark delays, thresholds, saturation, resource constraints, information gaps, incentives, and actors that adapt to the system.
- Check dimensional consistency, conservation where applicable, loop polarity, omitted exogenous drivers, boundary choices, and alternative structures that could produce the same observed behavior.
- Simulate or replay scenarios when the model can support it. Calibrate against observed history; otherwise label the model qualitative and do not invent parameter precision.
- Test interventions across short and long horizons. Look for burden shifting, compensating feedback, new bottlenecks, gaming, lock-in, and delayed harm.
- Recommend the smallest robust intervention set and the observations needed to validate or revise the model.
Output
Produce a Dynamic System Model:
- Question and behavior — decision, boundary, horizon, observations, and behavior-over-time patterns.
- Structure — stocks, flows, feedback loops, delays, constraints, nonlinearities, incentives, and adaptive actors.
- Model status — observed facts, hypotheses, parameters, calibration, competing structures, and confidence.
- Intervention analysis — leverage points, expected trajectories, counter-effects, failure modes, and rollback or containment.
- Validation plan — discriminating measures, scenario replay, monitoring, and conditions requiring model revision.
Boundaries
- Use
critical-analysisfor a bounded defect or competing root-cause hypotheses without material feedback dynamics. - Use
causal-inference-analysisto identify or estimate a specific intervention effect from experimental or observational evidence. - Use engineering architecture methods for a static dependency, module, or deployment graph.
- A causal-loop diagram is a hypothesis, not evidence that the represented links or intervention effects are true.
- Do not expand the boundary until it contains everything. Include a variable only when it could materially change the behavior or intervention.