Guide · Agent skill
How to use the Lean Startup Experiment Designer agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Lean Startup Experiment Designer is an agent skill, a packaged set of instructions that an AI agent such as Claude can load when a task matches it. This one, named brainstorm-experiments-new, helps you test whether people want a new product before you build it. This guide explains what the skill makes the agent do, how to install it, how to trigger it with plain requests, and where it falls short.
Lean Startup Experiment Designerphuryn/pm-skills · MIT Download the skill Quick start
- 1Download and unzip. Download the skill ZIP from this page and unzip it into ~/.claude/skills/ or your project's .claude/skills/ folder.
- 2Or upload it in the Claude apps. In the Claude apps, open Settings → Capabilities → Skills and upload the skill.
- 3Describe your product idea. Explain the new product, who it is for, and that you want to validate demand or create a pretotype.
- 4Attach any existing files. Add market research or mockups if you have them so the agent reads them before answering.
- 5Review the XYZ hypothesis. Check that the hypothesis names a specific target market and a clear behaviour, and challenge the percentage.
- 6Compare the proposed experiments. Look at the two or three suggested tests and pick the cheapest one that measures real commitment.
- 7Record the plan. Ask the agent to save the plan as markdown if it is long, so you can share it with stakeholders.
What is an agent skill and what does this one do?
An agent skill is a folder containing a SKILL.md file. The file holds instructions the agent follows when your request matches the skill's description. This skill comes from the phuryn/pm-skills project, under the pm-product-discovery section, and is released under the MIT licence.
When you ask the agent to validate a new product idea, the skill tells it to write an XYZ hypothesis in the form "At least X% of Y will do Z". X is the share of the target market expected to engage, Y is the specific target market, and Z is how they will engage. The agent then proposes two or three pretotypes, which are cheap tests that check real behaviour before a full product exists.
- Landing page: measures sign-ups or clicks.
- Explainer video: measures engagement as a sign of understanding and appeal.
- Email campaign: measures response and click-through rates.
- Pre-order or waitlist: measures willingness to pay.
- Concierge or manual MVP: delivers the service by hand to test whether it has value.
Install the skill with a ZIP file or the Claude apps
Download the ZIP file for this skill from this page and unzip it. You can install it for yourself or for a single project. For a personal install, unzip it into ~/.claude/skills/. For a project install, unzip it into the .claude/skills/ folder inside your project.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills. The skill package includes one file, SKILL.md.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills, then upload the skill
How do you trigger the skill with a natural request?
You do not need a special command. Describe the product idea and say you want to validate it, test demand, or create a pretotype. The skill's description covers requests about validating a new product idea, creating pretotypes, and testing market demand, so phrasing like that is the most reliable trigger.
Give the agent as much concrete detail as you can about the target customer and the product. If you have market research or mockups, attach them. The skill tells the agent to read any files you provide before answering.
- "Help me validate an idea for a meal-planning app for nurses."
- "Design pretotype experiments to test demand for a subscription bike-repair service."
- "I want to test market demand for a smart desk lamp before building it. Here is my landing page mockup."
What output should you expect?
For each experiment, the agent should name the hypothesis being tested, the method, the metric, and the success threshold. The skill also asks it to think step by step and to save the result as markdown if the write-up is substantial.
The skill is built around Alberto Savoia's principles from The Right It. These are skin-in-the-game, meaning tests of willingness to pay rather than interest alone, and Your Own Data, meaning your own experiments rather than market reports. The focus is on what people do, not what they say they think.
- One XYZ hypothesis for the idea.
- Two or three experiments, each with hypothesis, method, metric and success threshold.
- A markdown file if the plan is long enough to keep.
What does the skill need, and what about scripts and network access?
The skill lists no required dependencies. It does not run scripts and does not use the network, according to its published details. It works by reading the files you give it and writing its answer.
Be careful with the numbers it produces. The percentages and success thresholds are suggestions the agent makes. They are not researched benchmarks, so treat them as starting points to challenge, not as market data.
- Requires: nothing listed
- Runs scripts: no
- Uses network: no
Where does this skill fall short?
Do not use it for ideas that already have solid sales or usage data, because the framework targets unvalidated concepts. Avoid it for regulated or high-stakes launches. A fake landing page or pre-order could mislead customers or create legal problems, so check with counsel before running those tests.
The skill is also not a source of market figures. Use it to structure your thinking and to draft experiments, then collect your own data.
- Skip ideas that already have strong sales or usage data.
- Check with counsel before any test that takes payment or makes claims in a regulated market.
- Do not rely on its percentages as evidence.
What people use it for
- A founder testing demand for a new product with a landing page and waitlist before writing code.
- A product manager preparing a structured experiment plan to present to stakeholders.
- A student learning lean startup and pretotyping methods by applying them to a sample idea.
- A team choosing between a pre-order test and an explainer video for an unvalidated concept.
Frequently asked questions
How do I install the lean startup experiment agent skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for a personal install, or into .claude/skills/ inside a project. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
Does this skill require anything else to run?
The published details list no requirements. The skill runs no scripts and uses no network access, so it works by reading your files and writing its answer.
Is the lean startup experiment designer skill free?
The skill is released under the MIT licence, as listed on this page. Check the licence file in the source repository for the full terms.
What is an XYZ hypothesis?
It is a measurable statement in the form 'At least X% of Y will do Z'. X is the expected share, Y is the target market, and Z is the action they take. The skill asks the agent to write one for your idea.
Can the skill give me real market size or benchmark numbers?
No. The percentages and success thresholds are suggestions the agent makes, not researched benchmarks. Use them as starting points and gather your own data with the experiments.
What are the alternatives if this skill is not a fit?
The skill's own guidance is to rely on your own experiments rather than market reports, so for ideas with solid sales data you should use analytics you already have. For regulated launches, speak with counsel before running fake-demo tests.
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
Use this skill if you are a founder, product manager, student or team that wants a structured, cheap way to test demand before building. Pick something else if your idea already has real sales or usage data, or if you are launching in a regulated or high-stakes area where a fake landing page or pre-order could mislead customers. Treat its numbers as suggestions to test, not facts.
Download the skill