systems-thinking-skills

by BayramAnnakovVerified

Claude Code skills for systems thinking: hand-drawn stock-flow / causal-loop diagram critique, system archetype identification (Limits to Growth, Shifting the Burden, Fixes that Fail), and Meadows leverage point analysis.

14
Stars
3
Forks
HTML
Language
8/24/2026
Added
View on GitHubDownload ZIP

⚠️ Third-Party Software Notice

This skill is third-party open-source software developed and hosted independently on GitHub. SkillTip is an informational directory and does not control or maintain the underlying repository. Any security checks displayed are automated and limited in scope. Review the source code before installing.

Read the Terms of Service

Installation

Add to your Claude Code skills directory:

# Add to your Claude Code skills
git clone https://github.com/BayramAnnakov/systems-thinking-skills

Getting Started

Guides for using skills like systems-thinking-skills.

Security Report

Verified

Last scanned: —

{
  "status": "PASSED",
  "issues": []
}

README.md

Systems Thinking — Skills

Reusable Claude Code skills for systems thinking, drawn from the AI + Systems Thinking course (Empatika, Season 2026) by Bayram Annakov. Installable as a Claude Code plugin marketplace, also runnable as copy-paste mega-prompts in Codex / ChatGPT / Cursor.

The course is hands-on: participants draw stock-flow diagrams by hand, run live multiplayer simulations, and use these AI skills as a critic, builder, and amplifier — never as a designer. The skills enforce that pedagogy: they refuse to draw the diagram for you.

Available skills

/ai-systems-coach — Stock-flow diagram critic & archetype identifier (ships with W2)

You draw the diagram. The skill grades it.

Critiques a hand-drawn stock-flow / causal-loop diagram, identifies the system archetype (Limits to Growth, Shifting the Burden, Fixes that Fail), and surfaces Meadows leverage points. Outputs a Mermaid diagram, validates stocks, flags missing delays, prepares a clean stock-flow specification for Workshop 3 simulation.

Refuses to design a diagram from scratch, invent variables you didn't name, or force-fit an archetype.

Example:

/ai-systems-coach

Stock: Active Customers
Inflow: New signups (driven by referrals from Active Customers)
Outflow: Churn (driven by Service Quality, which drops past Capacity)
R: Active Customers → Referrals → New signups → Active Customers
B: Active Customers → Service Quality (delay 2 weeks) → Churn → Active Customers

/leverage-finder — Map your model to Meadows' 12 leverage points (ships with W4)

You bring a model and a goal. The skill returns 3 candidate interventions, ranked.

Takes a stock-flow model you have already built and a measurable goal you have already stated, then maps elements of your model to Donella Meadows' 12 leverage points and returns the 3 highest-leverage candidate interventions for your goal.

The skill enforces the asymmetry lesson structurally: at least one of the 3 candidates must be at knob ≤5 (rules, info flow, goals, paradigm). Most teams reach for parameters (knob 12) first because they have numbers to tune; the skill forces you to consider stronger leverage you can't see as easily.

Three structural rules:

  1. Phase 0 gate — no measurable goal, no output. The skill refuses to rank leverage without a stated purpose. "Improve the system" gets pushed back with three sharpening questions.
  2. No inventing structure — only maps stocks, flows, and parameters the model already contains. Won't add elements the user didn't draw.
  3. Forced choice — returns exactly 3 candidates, not a long list. Forcing the choice is part of the discipline.

Each candidate is formatted as: Meadows knob (number + name), what to change in the user's model, why it has leverage for the stated goal, one concrete experiment testable in 2 weeks, one caveat. Closes with an asymmetry paragraph noting which candidate is strongest and which feels most natural — those usually don't match.

Refuses: to invent stocks/flows/parameters not in the model, to write code, to march through all 12 leverage points as a checklist, to return more than 3 candidates.

/ai-stockflow-builder — Commission a runnable simulation app (ships with W3)

The skill emits a brief. Your AI agent (Claude Code or Codex) builds the app. You own the equations.

Walks you from a stock-flow diagram (or /ai-systems-coach output) to an interactive single-HTML-file simulation web app with sliders, chart, and equation chips — via three phases:

  1. Readback — skill reads your model back in plain English, surfaces ambiguity, forces ≥1 disagreement before any code
  2. Brief — skill emits a copy-paste commissioning brief; you paste it into Claude Code or Codex; your agent builds model.html with React via CDN, Recharts, Euler integration with dt=1
  3. Extraction test — you hand-compute t=1 from defaults and compare to the agent's app at t=1; mismatch = AI broke equations = stop and find where

If you don't have a diagram, Phase 0 — Interview mode asks Socratic questions and your answers become the model. The skill never proposes a stock or flow you didn't name.

Two-pass discipline:

PassCaps
Pass 1 (Simplified)exactly 1 stock, 2 flows, 2 parameters, ≤30 time steps, linear equations only — no min(), ternaries, or delays
Pass 2 (Full)no caps; extends Pass 1

When your model violates a Pass 1 cap, the skill runs a trim conversation — it proposes which entities to drop, demote to constants, or defer to Pass 2. Pass 1 forces "minimum viable model" discipline; Pass 2 layers complexity. Most workshops use Pass 1 first, then Pass 2 as a follow-up.

Refuses: to write the app itself (your agent does), invent missing parameters, deploy, stylize, or run the simulation.

/constraint-finder — Find the binding constraint (Theory of Constraints) (ships with W6)

You bring a flow and a goal. The skill finds the ONE wall — and whether it's a resource or a rule.

Takes a flow/process system and its goal, finds the single binding constraint, classifies it as a resource or (usually) a policy, and returns Goldratt's 5 Focusing Steps as concrete actions — Exploit and Subordinate before Elevate. This is the "find the wall" move.

The error it exists to kill: "find the slow thing and speed it up" — that's profiling, not ToC. The real constraint is almost never a machine; it's a rule, a metric, or an assumption.

Structural rules: a goal gate (no throughput unit, no output); a guard that refuses to force a single constraint on a demand-constrained or exploratory system; resource→policy promotion (default suspicion is a policy — far higher leverage than capacity); Exploit/Subordinate before Elevate (spending money first is the classic, costly ToC error); and Subordinate kept as the climax (fast resources must be willing to sit idle).

Refuses: to optimize a non-constraint, to Elevate before Exploit/Subordinate, to proceed without a goal, to call a policy constraint a resource, or to invent steps.

/triz-dissolve — Dissolve a trade-off (TRIZ) (ships with W6)

You bring a trade-off. The skill eliminates the contradiction — it doesn't balance it.

Takes a trade-off ("improving X worsens Y") and dissolves it with TRIZ: reframe as a physical contradiction, state the Ideal Final Result, then scan the full 40 inventive principles + the 4 separation principles + resource analysis to generate genuinely different structural resolutions — each run through a dissolve-vs-relocate test. This is the "move the wall" move.

The LLM already knows TRIZ, so the skill doesn't re-teach it. It does two things: enforces the discipline the LLM skips (left alone it defaults to compromise and lets cost-relocations pass as solutions), and directs it to use the full toolkit ("you ARE the contradiction matrix" — reason over all 40 principles, don't cite hallucination-prone matrix cells).

Structural rules: refuse compromise ("find the optimal balance" is never an answer); a structure-vs-physics guard (don't fake a dissolution for a real law — optimize the honest compromise instead); and a dissolve-vs-relocate test on every resolution (did the cost leave the system, or just move to another team / a later time / tech debt?).

Refuses: to return a compromise, to present a relocation as a dissolution, to fake-dissolve a real physical law, or to name a principle without showing how it separates the demands.

The W6 pair is designed to compose: /constraint-finder locates the wall, /triz-dissolve moves it.

/why-tree — Build the evidence-graded Current-Reality Tree (power instrument)

You bring a hard, contested problem. The skill builds the evidence and finds where the wall even is.

Where /constraint-finder takes a system you've already mapped, /why-tree is the heavyweight, multi-agent instrument for a messy problem you haven't — it fans 20-55 AI agents across the branch-space, grades every node by the kind of evidence behind it (MEASURED / INFERENCE / CLAIM / HYPOTHESIS…), tries to refute its own load-bearing branches, and converges on the ONE constraint — plus the negative branches (fixes that backfire) and the single cheapest test that would fork-decide. Output: a self-contained interactive HTML tree + a one-page decision doc. The natural chain is /why-tree/constraint-finder: this locates and evidences the constraint; constraint-finder gives the focusing steps.

Unlike the course skills above, this one does NOT refuse to do the thinking — it's a power tool, token-heavy by design (Standard ~20-36 agents; Deep ~30-55) and it requires Claude Code's multi-agent Workflow tool (the included PROMPT.md is a degraded single-context fallback). It gates on a depth + token-cost choice before launching, and it tells you when not to reach for it: a clear-cut problem a single careful pass would nail — just ask one agent.

Structural rules: grade every node (no ungraded assertions); refute before trust (a degraded/zeroed refute pass is loudly flagged); converge to ONE constraint; the apex is a declarative statement, never a question or a smuggled solution; an honest census ("a map, not a verdict" when the decisive nodes are unmeasured); the constraint branch drilled to bedrock as a multi-level chain.

Refuses: to ship five co-equal roots, to delete refuted branches, to run the full workflow on a problem one agent would nail, or to state a confident verdict on HYPOTHESIS-graded nodes without disclosing it.

Installation

The skills follow the agentskills.io open standard — installable in both Claude Code and OpenAI Codex with the same SKILL.md files.

Option 1: Claude Code plugin install

/plugin marketplace add BayramAnnakov/systems-thinking-skills
/plugin install ai-systems-coach@systems-thinking-skills
/plugin install ai-stockflow-builder@systems-thinking-skills
/plugin install leverage-finder@systems-thinking-skills
/plugin install constraint-finder@systems-thinking-skills
/plugin install triz-dissolve@systems-thinking-skills
/plugin install why-tree@systems-thinking-skills
/reload-plugins

Option 2: Codex CLI

git clone https://github.com/BayramAnnakov/systems-thinking-skills.git ~/.agents/systems-thinking-skills
mkdir -p ~/.agents/skills
ln -s ~/.agents/systems-thinking-skills/skills/ai-systems-coach ~/.agents/skills/ai-systems-coach
ln -s ~/.agents/systems-thinking-skills/skills/ai-stockflow-builder ~/.agents/skills/ai-stockflow-builder
ln -s ~/.agents/systems-thinking-skills/skills/leverage-finder ~/.agents/skills/leverage-finder
ln -s ~/.agents/systems-thinking-skills/skills/constraint-finder ~/.agents/skills/constraint-finder
ln -s ~/.agents/systems-thinking-skills/skills/triz-dissolve ~/.agents/skills/triz-dissolve
ln -s ~/.agents/systems-thinking-skills/skills/why-tree ~/.agents/skills/why-tree

Codex auto-discovers skills under ~/.agents/skills/. See Codex skills docs.

Option 3: Clone + copy (any runtime)

git clone https://github.com/BayramAnnakov/systems-thinking-skills.git
# Claude Code:
cp -r systems-thinking-skills/skills/ai-systems-coach ~/.claude/skills/
cp -r systems-thinking-skills/skills/ai-stockflow-builder ~/.claude/skills/
cp -r systems-thinking-skills/skills/leverage-finder ~/.claude/skills/
cp -r systems-thinking-skills/skills/constraint-finder ~/.claude/skills/
cp -r systems-thinking-skills/skills/triz-dissolve ~/.claude/skills/
cp -r systems-thinking-skills/skills/why-tree ~/.claude/skills/
# Or Codex:
cp -r systems-thinking-skills/skills/ai-systems-coach ~/.agents/skills/
cp -r systems-thinking-skills/skills/ai-stockflow-builder ~/.agents/skills/
cp -r systems-thinking-skills/skills/leverage-finder ~/.agents/skills/
cp -r systems-thinking-skills/skills/constraint-finder ~/.agents/skills/
cp -r systems-thinking-skills/skills/triz-dissolve ~/.agents/skills/
cp -r systems-thinking-skills/skills/why-tree ~/.agents/skills/

Option 4: Mega-prompt (no skills runtime — plain ChatGPT, Cursor, Claude.ai)

If your runtime doesn't support skills, copy the PROMPT.md from each skill's directory. Paste the block between ===PROMPT START=== and ===PROMPT END=== into your tool, then paste your model.

Usage

/ai-systems-coach          # paste your hand-drawn diagram in structured form
/ai-stockflow-builder      # paste your model, or say "interview me" for Phase 0
/leverage-finder           # paste your model + measurable goal; get 3 ranked leverage candidates
/constraint-finder         # paste your flow + goal; find the binding constraint + the 5 focusing steps
/triz-dissolve             # paste a trade-off ("improving X worsens Y"); dissolve the contradiction

The W2 → W3 progression: run /ai-systems-coach after Workshop 2 to grade your archetype homework. Take its "Simulation prep" output and paste it into /ai-stockflow-builder after Workshop 3 to build the runnable app.

Both skills respond bilingually — they auto-detect English / Russian and use whichever you typed in.

Course context

Part of the AI + Systems Thinking course (6 biweekly online sessions, 2026):

  1. How Systems Work — stocks, flows, feedback, delays
  2. System Archetypes — Limits to Growth, Shifting the Burden, Fixes that Fail (ships /ai-systems-coach)
  3. From Diagram to Simulation — turning your diagram into runnable code (ships /ai-stockflow-builder)
  4. Goal of the Model + Meadows Leverage — from "I have a model" to "I know which knobs matter" (ships /leverage-finder)
  5. Optimization + MCP — connecting models to real data
  6. Theory of Constraints + TRIZ — find the binding constraint, then dissolve the contradiction (ships /constraint-finder + /triz-dissolve)

More skills will land in this repo as later workshops ship.

Pedagogical philosophy

These skills enforce a strict separation: humans draw and own equations; AI critiques and builds infrastructure. We don't ship AI tools for skills participants haven't first practiced manually. That's why:

  • /ai-systems-coach ships with W2 (after archetypes are taught by hand) and refuses to design a diagram from scratch
  • /ai-stockflow-builder ships with W3 (after stock-flow notation is taught) and refuses to invent equations the human didn't name
  • /leverage-finder ships with W4 (after the goal-of-the-model and Meadows asymmetry are taught on cohort cases) and refuses to rank leverage without an explicit, measurable goal
  • /constraint-finder and /triz-dissolve ship with W6 (after ToC and TRIZ are taught and practiced on a live commons game) and refuse, respectively, to optimize a non-constraint and to return a compromise

The skills also encode the workshop's force-disagreement mechanic. Both refuse to proceed if the user says "looks perfect" — they push back with units, polarity, periodicity, or threshold-vs-linear questions until the user has corrected something. Drawing and disagreeing are the act of thinking.

If you want to design from scratch, draw it on paper first. The friction is the lesson.

License

MIT — see LICENSE.

Frequently Asked Questions

What is systems-thinking-skills?

systems-thinking-skills is an open-source ide extensions skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by BayramAnnakov. Claude Code skills for systems thinking: hand-drawn stock-flow / causal-loop diagram critique, system archetype identification (Limits to Growth, Shifting the Burden, Fixes that Fail), and Meadows leverage point analysis. It has 14 GitHub stars.

Is systems-thinking-skills safe to use?

Yes. systems-thinking-skills passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.

How do I install systems-thinking-skills?

Clone the repository with "git clone https://github.com/BayramAnnakov/systems-thinking-skills" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is systems-thinking-skills written in?

systems-thinking-skills is primarily written in HTML. It is open-source under BayramAnnakov on GitHub, so you can review or fork the full source.

Are there alternatives to systems-thinking-skills?

Yes. SkillsLLM lists many other IDE Extensions skills you can browse and compare side by side. Open the IDE Extensions category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh systems-thinking-skills against similar tools.

Comments (0)

No comments yet. Be the first to share your thoughts!

claudian

by YishenTu

An Obsidian plugin that embeds Claude Code/Codex as an AI collaborator in your vault

14,921972TypeScript
IDE Extensions
View details

OpenMythos

by kyegomez

A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.

14,8003,292Python
IDE Extensions
View details

cursor-hooks

by cursor

5

Automate Cursor behavior around agent events with hooks.

12,500890TypeScript
IDE Extensionsclaude-codeai-tools
View details

arscontexta

by agenticnotetaking

Claude Code plugin that generates individualized knowledge systems from conversation. You describe how you think and work, have a conversation and get a complete second brain as markdown files you own.

3,479220Shell
IDE Extensions
View details

Developers Also Liked

Based on votes and bookmarks from developers who liked this skill

ECC

by affaan-m

10

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

242,21936,702JavaScript
AI Agentsai-agentsanthropicclaude-code
View details
15

An agentic skills framework & software development methodology that works.

234,96620,863Shell
AI Agentsai-agentsbrainstorming
View details

n8n

by n8n-io

12

Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.

201,88160,308TypeScript
MCP Serversapisai-tools
View details

The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.

185,94028,768JavaScript
AI Agentsai-agentsanthropicclaude-code
View details

cc-switch

by farion1231

3

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io

128,8688,826Rust
AI Agentsclaude-codeai-tools
View details