Guide · Agent skill
How to use the Systems Thinking Analyst agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
SystemsThinking is an agent skill that analyses a recurring problem by looking for the structure that keeps producing it, instead of blaming the last event or a person. It is built from the LifeOS project, released under the MIT licence, and is aimed mainly at people who run an AI assistant setup and want to know why a metric, habit or signal keeps drifting. After this guide you will know what the skill does, how to install it, how to ask for it in plain language, and where it is a poor fit.
Systems Thinking Analystdanielmiessler/LifeOS · MIT Download the skill Quick start
- 1Download the ZIP. Download the SystemsThinking ZIP from this page.
- 2Unzip it into a skills folder. Unzip into ~/.claude/skills/ for all projects, or into .claude/skills/ inside one project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, then Capabilities, then Skills, and upload the skill there.
- 4Describe a recurring problem. Ask why the problem keeps happening and include the measured facts you already have.
- 5Review the diagram and the ranking. Check the loops, the evidence and the inferred links the output marks, then review the leverage point it recommends.
What an agent skill is and what SystemsThinking adds
An agent skill is a folder of instructions and reference files that an AI agent loads when a request matches the skill's purpose. The agent then follows the method in those files rather than improvising one.
SystemsThinking covers five kinds of structural analysis: the Iceberg model (walking from symptoms down to causes), Causal Loop feedback diagrams, archetype matching, Meadows leverage points, and concept maps. Its premise is that behavior is generated by structure: loops, delays, incentives, flows and beliefs.
- Iceberg: moves from visible events to patterns, then to the structure underneath.
- CausalLoop: builds feedback loop diagrams and labels each loop as reinforcing or balancing.
- FindArchetype: matches a situation to known patterns such as Shifting the Burden, Fixes That Fail, Limits to Growth and Tragedy of the Commons.
- FindLeverage: ranks where a change would have the most effect.
- ConceptMap: maps entities and the relationships between them.
Install the SystemsThinking skill from a ZIP file
Download the SystemsThinking ZIP from this page, then unzip it. You can install it in one of two places. For all your projects, unzip it into the skills folder in your home directory, ~/.claude/skills/. For a single project, unzip it into a .claude/skills/ folder inside that project.
If you use the Claude apps instead of Claude Code, you can upload the skill under Settings, then Capabilities, then Skills. Use the upload option there rather than placing files on disk.
- Global install: ~/.claude/skills/
- Per-project install: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills
~/.claude/skills/How to trigger the skill with a natural request
You do not need to type a command. Describe a recurring problem and ask why it keeps happening. The skill's trigger words include systems thinking, causal loop, feedback loops, archetypes, leverage points, iceberg model, fix the system, why does this keep happening, recurring problem and second-order effects.
The analysis works best when you include the facts you already have, such as counts, rates and dates. The skill is designed to start from the facts a request supplies, so a request with measured numbers gives it something concrete to test.
- Include the measured facts up front, for example the rate before and after a change.
- Name the system: your assistant setup, a dashboard, a team process or a real-world system.
- Say what outcome you want to change, so the leverage ranking has a target.
Examples of requests that fit
The skill's own examples show the pattern. A request about a model router that delegates less after a model swap would trigger CausalLoop and then FindLeverage. A question about why a skill library keeps growing would trigger Iceberg, FindArchetype and FindLeverage. A question about a city that keeps running short of water after each reservoir expansion would trigger CausalLoop, FindArchetype and FindLeverage.
A dashboard metric that has shown the same frozen value for months is another fit. The skill would look for the independent failures behind the frozen value and a leverage point that catches the whole class of silent signal loss.
- Metric drifting the wrong way in your AI setup
- A library or set of hooks that keeps growing
- A signal that died without anyone noticing
- A real-world system such as a city's water supply or a team's on-call rotation
What the skill needs and what it does not do
The skill lists no required tools or prerequisites. It does not run scripts and does not use the network, according to its listing. It works on files and information that you give it and that it reads from paths it is allowed to access.
The skill is not for live incidents or outages. Its own guidance says to stabilise first and analyse afterward, and that incident causal chains belong to a separate RootCauseAnalysis skill. It is also not for quick factual questions, because its method is slow and structural and can over-theorise simple problems.
- No listed requirements
- Runs scripts: no
- Uses network: no
- Not for live incidents: use RootCauseAnalysis for incident causal chains
- Not for simple factual questions
Customise the skill with your own preferences
You can override the defaults by placing files in a customisation folder. The location is based on the CLAUDE_CONFIG_DIR environment variable when it is set, and otherwise on ~/.claude. If the folder does not exist, the skill uses its built-in defaults.
The skill also keeps an optional History.md file in that folder that records past runs and their measured facts. According to the skill's documentation, nothing reads that file at run time.
- Preferences, configurations or resources in the folder override the defaults.
- If the folder is missing, the defaults apply without error.
- History.md is a record only and is not read during runs.
<configRoot>/LIFEOS/USER/CUSTOMIZATIONS/SKILLS/SystemsThinking/What people use it for
- Finding why a metric keeps drifting the wrong way in an AI assistant setup.
- Mapping feedback loops behind a habit or process that keeps recurring.
- Checking whether a recurring problem matches a known archetype such as Shifting the Burden.
- Ranking which change would have the most effect on a recurring problem.
- Learning systems thinking with structured diagrams and archetype checks.
Frequently asked questions
Is the SystemsThinking skill free to use?
The skill is released under the MIT licence, as listed on this page. Check the LifeOS repository for the full licence text before you reuse or redistribute it.
What do I need installed to run SystemsThinking?
The listing shows no required tools. The skill does not run scripts and does not use the network, so it needs only an agent that can load skills.
Where do I put the skill folder after downloading?
Unzip it into ~/.claude/skills/ to make it available everywhere, or into .claude/skills/ inside one project. In the Claude apps, you can instead upload it under Settings, Capabilities, Skills.
Why isn't my skill triggering when I ask about a problem?
The skill triggers on words such as systems thinking, feedback loops, archetypes, leverage points, iceberg model, and why does this keep happening. Use one of these phrases and describe a recurring problem rather than a one-off error.
Should I use SystemsThinking or RootCauseAnalysis?
Use SystemsThinking for recurring behavior and structural causes. The skill says incident causal chains belong to RootCauseAnalysis, and during a live incident you should stabilise first.
Does SystemsThinking work outside software and AI setups?
Yes. The skill can analyse real-world systems, such as a city's water supply or a team's on-call rotation, though it is designed mainly for a solo operator's AI setup.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
SystemsThinking suits people who keep meeting the same problem and want the structure behind it mapped with loops, archetypes and ranked leverage points, including operators of AI assistant setups and students learning the method. Skip it for live outages, where you need to stabilise first and use RootCauseAnalysis, and for simple factual questions, where its slow method will over-theorise.
Download the skill