How to use the Growth Loops Analyzer agent skill

What the Growth Loops skill makes the agent do

Growth loops (also called flywheels) are repeating cycles in which the output of using a product brings in new users. The skill asks the agent to evaluate five loop types: Viral, Usage, Collaboration, User-Generated and Referral. For each one it assesses how well the loop fits your product.

Once you describe your product, the agent ranks the five loop types, recommends a primary loop and lists secondary loops to layer on later. It also suggests key metrics and a 30-60-90 day roadmap aimed at self-sustaining acquisition rather than relying on paid ads. The skill's stated purpose is to reduce reliance on paid acquisition and build product-led growth.

• Viral: users create content that is shared on external platforms and brings in new users.
• Usage: users create or share something inside the product that invites others or drives re-engagement.
• Collaboration: users invite colleagues to work together, spreading the product inside organizations.
• User-Generated: users discover content made by others, then create and share their own.
• Referral: users invite others in exchange for rewards, incentives or social recognition.

Install the Growth Loops skill

Download the growth-loops ZIP file from the skill's page in this directory. The skill is a folder that contains a single file, SKILL.md. Unzip it so the growth-loops folder sits in one of the skill locations below.

You can install it for yourself across all projects by placing it in your personal skills folder, or for one project only by placing it in that project's skills folder. If you use the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

• Personal install: put the growth-loops folder in ~/.claude/skills/
• Project install: put the growth-loops folder in the .claude/skills/ folder of your project
• Claude apps: go to Settings → Capabilities → Skills and upload the ZIP

How to trigger the skill with natural requests

You do not need a special command. Describe what you are trying to do in plain language, and the agent can match your request to the skill's description, which covers designing growth mechanisms, building product-led traction and understanding how growth loops work.

The more context you include, the more useful the ranking will be. The skill's input guidance lists the product description and primary user action, target users and their behaviour, existing sharing or collaboration features, current growth channels and metrics, and any constraints.

• Which growth loop should my project management app build first? Here is what it does and who uses it: ...
• Analyze how a note-taking tool could grow through sharing. Our current channels are paid search and word of mouth.
• Rank the five growth loop types for my design tool and give me a 30-60-90 day roadmap.
• Explain how a referral loop would work for our budgeting app and what would make it expensive.

Example of a useful request and what to expect back

A good request names the product, the action users take most, who the users are, and any sharing or team features you already have. The output then follows the skill's structure: a ranked evaluation of the five loop types, a recommended primary loop with an implementation plan, secondary loops, key metrics and a measurement framework, a 30-60-90 day roadmap, and an estimate of loop coefficient and growth.

Treat the ranking and roadmap as a structured starting point for discussion with your team. The skill works from what you tell it in the conversation; it does not look up your data or market figures.

What the skill needs and how it runs

The skill lists no requirements. It runs no scripts and makes no network requests. The package contains only SKILL.md, which is text instructions, with no code or data sources of its own.

Because nothing is fetched automatically, all the facts the agent uses must come from your message. Be careful about sharing confidential product or customer details, since they will be part of the conversation you are having with the agent.

• No listed requirements
• No scripts included or run
• No network use
• Only file in the package: SKILL.md
• Licence: MIT

Tips for getting reliable results

Start with one loop and master it before adding complexity, as the skill's own guidance suggests. Use the skill to choose where to focus, then test the chosen loop with real usage data.

The skill asks for projections and loop coefficients but has no data source. Check every number the agent produces against your own analytics before you use it in a plan or a forecast.

• Give concrete details about how users share or invite others today.
• Ask the agent to list the assumptions behind any growth estimate, then verify them.
• Measure loop health weekly during the optimisation phase.
• Combine loops only after the first loop is operating at scale.