Test It Before You Launch It
Three free Founder Skill checks on your rivals, your numbers and 100 simulated buyers, before any stock is ordered.

This guide is for people about to launch a physical or online product who want answers before they buy stock. Founder Skill is a free set of 11 Claude skills under the MIT licence. You install it in Claude Code, then run three: competitors, which lists rivals, their prices and customer complaints with links; CFO, which shows what each sale leaves after its costs, the sales a day needed to break even and the cash needed in year one; and consumer, which shows the product to 100 simulated buyers and lists who buys, who passes and why. Each prompt is below with brackets for your own details, and each result has a check you can do by hand. The buyers are simulated, so read the buy rate as a ceiling, and a full run uses your Claude plan. You decide, and you confirm with real people before ordering stock.
The test most products never get
A new product usually gets tested by launching it. The price is set and the stock is bought before anyone has said whether they would buy. The expensive mistakes are made in that order: a price that leaves too little after costs, stock ordered for a daily sales rate that never arrives, a product buyers pass on for a reason that was easy to hear early.
This guide runs three checks before any money goes out, using Founder Skill, a free set of skills for Claude. The first lists your rivals, their prices and their customers' complaints. The second works out what each sale leaves after its costs and how many sales a day you need to break even. The third shows your product to 100 simulated buyers and lists who buys, who passes and why.
In the reel the product is the Fernway Trail Harness, an invented no-pull dog harness at $34. Every number in this guide that mentions it is made up for the example.
Completion test: you end with three files in a folder called founder: competitors.md, cfo.md and panel/results.md. Open each one. You can say who else sells this and at what price, how many sales a day break even needs, and who passes and why. If you cannot, the check is not finished.
What Founder Skill is, in plain words
Founder Skill is a free set of 11 skills for Claude, made by Jake Schincariol and released under the MIT licence. Its README says there is no signup, no API key and nothing to connect. A skill is a folder of instructions that Claude follows when you ask for it by name.
This guide uses three of the 11. The other eight cover a board vote on your idea, pricing, offers, marketing, branding, operations, a launch plan and a final business plan.
Every skill saves its work in a folder called founder inside your project, and later skills read what earlier ones wrote. That is why you can run them one after another without retyping your idea.
| Skill | Command | You give it | It writes |
|---|---|---|---|
| Competitors | /founder-competitors | What you sell, who buys it, your price | founder/competitors.md and founder/competitors.csv |
| CFO | /founder-cfo | Price, cost of one sale, monthly costs, startup costs, daily sales plan | founder/cfo.md, founder/numbers.json and founder/cfo-sources.md |
| Consumer | /founder-consumer | A target customer and a short pitch | A founder/panel folder with results.md |
If you install it as a Claude Code plugin, the commands carry a prefix, for example /founder-skill:founder-cfo. Asking in plain words, as the prompts below do, works either way.
What you need and how to set it up
| Need | What it means |
|---|---|
| Claude Code | Anthropic's coding tool. It runs in a terminal, in the Claude desktop app (the Code tab), in VS Code and JetBrains, and in the browser. The consumer skill needs it, because it starts one helper per buyer |
| A Claude account | Most Claude Code surfaces need a Claude subscription or an Anthropic Console account. The desktop app needs a paid subscription |
| Python 3.8 or newer | The CFO and consumer skills use small Python programs. The README says no extra installs are needed |
| A project folder | An empty folder on your computer. The founder folder is created inside it |
| Web access for Claude | The competitors skill searches the web and reads pages |
| One decision-maker | The person who reads the three results and decides whether to launch |
What it costs. The skills are free. Running them uses your Claude plan. The competitors skill reads many web pages, and a full consumer run is 100 helper calls in about 10 waves, plus any retries. Let it run a 20-buyer quick pass first if it offers one.
Data. What you type, including your price and costs, goes to Anthropic like any Claude chat.
Install Claude Code (skip this if you already have it). On macOS, Linux or WSL, run:
curl -fsSL https://claude.ai/install.sh | bashOn Windows PowerShell, run:
irm https://claude.ai/install.ps1 | iexOr download the desktop app from Anthropic's Claude Code page and open its Code tab. Then check it. Open a new terminal window and run:
claude --versionPass: a version number prints.
Stop: if the command is not found, the install folder is not on your PATH yet. Anthropic's installation troubleshooting page has the fix.
Install Founder Skill. Open your empty project folder in Claude Code. In the terminal, go to the folder and start it:
cd your-project
claudeThen run these two lines inside Claude Code:
/plugin marketplace add Jakeschincariol/founder-skill
/plugin install founder-skill@founder-skillTwo other routes from the README give the same skills:
- Paste the link. Paste https://github.com/Jakeschincariol/founder-skill into Claude and say install skill.
- Copy the folders. Run
git clone https://github.com/Jakeschincariol/founder-skill.gitand thencp -r founder-skill/skills/founder-* ~/.claude/skills/.
The README also says you can paste a single SKILL.md into any Claude chat to run it as a mode. That route loses the helpers and the Python tools, so the 100-buyer panel does not run that way and the CFO maths is done by the model instead of the program.
Checkpoint. Type /founder in Claude Code. The founder skills appear in the list. Pass: you can see founder-competitors, founder-cfo and founder-consumer. Stop: nothing appears, so restart Claude Code and repeat the install.
Write down your details first
Each prompt below has brackets to fill. Ten minutes with your own quotes and invoices beats any guess. The CFO skill ranks your quotes and invoices first, public prices second and its own estimates last, and it labels the estimates.
| Skill | What to have ready |
|---|---|
| Competitors | The product and who buys it, your price, where you sell, how it ships, how you promote it, your startup budget, your monthly running costs, your team and who makes the product |
| CFO | Your price; the cost of one sale split into parts (the product, packaging, shipping, card fees); monthly costs; startup costs; units sold per day for each of the first 12 months; the daily sales you plan once it is running; the most you can handle in a day |
| Consumer | Your target customer; 2 to 4 segments with a share and an income range; 4 to 8 buying behaviours with shares; 4 to 8 objections; and 3 to 6 plain lines on what every buyer sees |
Two rules keep the results usable. Segment shares add up to 100%, and behaviour shares add up to 100%. Prices are before tax, so note whether sales tax is added on top where you sell.
Skill 1: competitors
Open Claude Code in your project folder and send this as one message. Replace the bracketed parts and keep the rest.
The rest of this playbook is below.
Enter your email to read it. You also get one new playbook every week.
One email a week. Unsubscribe any time.
Run the founder competitors skill. I'm launching [product name], [what it is] for [who buys it: age range, country, what they buy and how often]. It sells for $[price], only [where you sell], shipped [where and how long it takes], and promoted with [how you promote it]. The startup budget is $[startup budget] and running costs are $[monthly running costs] a month. It's [your team], and [who makes the product].In the reel the filled-in version is about the Fernway Trail Harness: a $34 adjustable no-pull dog harness for US dog owners aged 25 to 55, sold only on the founder's own website, a $7,750 startup budget and $4,400 a month in running costs.
What you get. The skill saves founder/competitors.md and founder/competitors.csv. The analysis has six parts:
- A competitor table, sorted by how directly each one competes
- The price range for the comparable item: lowest, median and highest
- A text map of where each one sits, on two axes that matter to the buyer
- Customer complaints, ranked, with links and counts
- The gap: what nobody selling this does well that your product could own, tied to evidence, or a plain statement that no gap exists
- The threat: who could copy you fastest, and how
It looks at 5 to 10 direct and indirect competitors plus the main substitutes, which are the ways buyers solve the problem without buying from a business at all. It reads 1 to 3 star reviews first. A complaint that appears in fewer than three reviews is marked thin.
Check it before you use it.
| Check | How |
|---|---|
| The links work | Open three competitor links from the table and confirm the product and the price on the page |
| Same product | Each price must be for the closest comparable item to yours, not a bundle or a different size |
| Complaints are real | Pick one ranked complaint and open the review behind it |
| The gap has evidence | The gap must point to complaints or prices in the table, or say no gap exists |
Pass: the three links open to what the table says, and the gap is tied to something you can see.
Stop: a link that does not open, a price that is not on the page, or a gap with no evidence. Ask the skill to remove it or to say it could not verify it.
Skill 2: CFO
Send this in the same project. Put the cost of one sale in parts, so each part can be checked. Card fees belong in the cost of one sale.
Run the founder CFO skill for [product name], [what it is] sold [where], with these numbers.
One sale is one [unit]. The price is $[price].
The cost of one [unit]: [part] $[amount], [part] $[amount], [part] $[amount], card fees ([percent]% + $[fixed fee]) $[amount].
Monthly costs: [item] $[amount], [item] $[amount], [item] $[amount], [item] $[amount].
Startup costs: [item] $[amount], [item] $[amount], [item] $[amount], [item] $[amount].
[Units] sold per day in each of the first 12 months: [month 1], [month 2], [month 3], [month 4], [month 5], [month 6], [month 7], [month 8], [month 9], [month 10], [month 11], [month 12]. Once it is running, the plan is [number] a day. The most we can pack and ship in a day is [number]. Use 30 days a month.The skill turns these into an input file, founder/numbers.json, with a price, a cost per sale, monthly costs, startup costs, a daily plan, a 12-month ramp and a daily capacity. If you leave out a number only you know, it asks rather than guessing.
What you get. founder/cfo.md opens with a short CFO's note, then shows:
- Contribution per sale: what each sale leaves after its own costs
- Break-even, in sales per day
- Profit margin at your plan volume
- Year one, month by month
- Year-one operating profit, and the result after startup spend
- The cash needed before the business pays for itself
- The month the startup spend is earned back
- What happens if the price drops 10%, volume drops 20% or unit costs rise 15%
- Flags: a price below the cost of one sale, break-even above your daily capacity, a loss in year one, and payback beyond 12 months
Check the maths by hand. The skill runs a Python program, so the figures should match a calculator. The example below uses the reel's invented harness numbers.
| Figure | How to work it out | Fernway example |
|---|---|---|
| Cost of one sale | Harness $9.80 + packaging $1.10 + packing and shipping $5.20 + card fees $1.29 | 17.39 |
| Contribution per sale | Price minus the cost of one sale | $34.00 - $17.39 = $16.61 (48.9% of the price) |
| Monthly costs | Ads $2,800 + packer $1,200 + platform $140 + insurance and accounting $260 | 4400 |
| Break-even per month | Monthly costs divided by contribution | 4,400 / 16.61 = 265 sales |
| Break-even per day | Break-even per month divided by 30 | 8.8, so 9 a day |
| Month 1 | 5 a day x 30 = 150 sales x $16.61, minus $4,400 | -$1,908.50 |
| A month at the plan of 15 a day | 450 sales x $16.61, minus $4,400 | +$3,074.50 |
| Startup costs | $3,200 + $900 + $2,200 + $700 + $750 | 7750 |
Worked month by month, the example loses money in months 1 to 3 and first earns a profit in month 4. The running total, counting the $7,750 startup spend, is lowest in month 3 at about -$11,482 and turns positive in month 11. Year-one operating profit is about $12,976, or about $5,226 after startup costs.
That low point is the line that can save you money. The $7,750 budget is all spent on startup costs, so in this example about $3,732 of early losses has to come from somewhere else. The skill states its own cash needed figure, and its definition can differ slightly from this hand count, so compare the shape (the low month and the month it turns positive) rather than the last dollar.
Pass: your hand figures for contribution and break-even match the skill to the cent, and its cash needed is not smaller than your startup spend.
Stop: a contribution or break-even that does not match, a number you did not enter, or a flag you did not expect. Ask the skill which input produced the number.
Skill 3: consumer
The consumer skill needs Claude Code, because it starts one helper per simulated buyer. Send this in the same project. The skill shows you the customer profile it built from your message before it runs, so read that first.
Run the founder consumer skill with 100 buyers.
Business: [business name], [what it sells and where it ships].
Target customer: [who they are: age range, country, what they buy and how often].
Segments: [segment 1] [share]% (income $[low] to $[high]), [segment 2] [share]% (income $[low] to $[high]), [segment 3] [share]% (income $[low] to $[high]).
Buying behaviours:
- [behaviour 1], [share]%: [one line on how they buy].
- [behaviour 2], [share]%: [one line on how they buy].
- [behaviour 3], [share]%: [one line on how they buy].
- [behaviour 4], [share]%: [one line on how they buy].
Objections buyers have:
- [objection 1]
- [objection 2]
- [objection 3]
- [objection 4]
Everyday habits: [what they do each day that matters to the purchase].
What every buyer sees:
[product name], [one line on what it is].
[Features in one line].
$[price].
[Where it is sold and how long delivery takes].
[Returns policy].The skill takes 2 to 4 segments, 4 to 8 buying behaviours and 4 to 8 objections, and the pitch is 3 to 6 plain lines. Every buyer sees the same pitch. Real customer data or the complaints from your competitors result are better inputs than guesses. In the reel the business is Fernway Pet and the harness is the $34 Trail Harness.
What you get. A founder/panel folder holds the buyer cards, their answers and results.md. The results show:
- The headline: how many buy and how many pass, for example 40 buy and 60 pass
- Which segments and buying behaviours buy most and least
- The top three reasons buyers pass, each with one quote
- The answers that came up most about what would turn a no into a yes
- One change to make before spending money, and which skill to use for it
Each buyer also says whether they would buy in the first month and how they react to four prices: too cheap, a bargain, getting expensive and too expensive.
What a full run involves. The skill launches the buyers in waves of 10 and re-briefs any buyer whose answer is missing or invalid. A buyer that fails twice is left out and reported. In the terminal, each buyer saves a file, so Claude Code asks permission each time unless you switch to accept-edits mode with Shift+Tab. If Claude offers the 20-buyer quick pass, take it first.
Read the result as a ceiling. The buyers are simulated, and language models tend to agree. The README says to read the buy rate as an upper bound and to confirm any big decision with real people.
| Check | How |
|---|---|
| The profile is yours | Before the run, confirm the segments, shares and objections match your real customers |
| The reasons are real | Open founder/panel and read five buyer answers behind the top reason for passing |
| The quotes are from buyers | Each quote in results.md must appear in a buyer's answer file |
| One change at a time | When you test a new price or offer, change only that and run the panel again |
Pass: the top reasons for passing match objections you already hear or can see in the competitor complaints.
Stop: a headline buy rate you would only believe on a good day, or reasons that have nothing to do with your product. Fix the profile and run again.
Put the three answers together
Each skill answers one question. Read them side by side before you order stock.
| Question | Where the answer is | What should make you stop or change something |
|---|---|---|
| Is there room for it? | founder/competitors.md: price range and the gap | Your price is above the highest comparable price with no reason a buyer would pay it, or the skill says no gap exists |
| Do the numbers work? | founder/cfo.md: break-even, payback, cash needed, flags | Break-even above what you can ship in a day, a loss in year one, payback past 12 months, or cash needed above your startup budget |
| Will buyers say yes? | founder/panel/results.md: buy and pass, reasons, what flips a no | A buy rate you would not bet on, or a top reason for passing you cannot fix |
Three outcomes are possible. Go ahead as planned, change the price, offer or costs and run the affected skill again, or stop before any money goes out. Whatever you pick, confirm it with real people before the first stock order: a waitlist, a pre-sale or a priced landing page. The simulated buyers only tell you what to test.
The results connect. If the buyers object to price, the competitors file shows what rivals charge and the CFO file shows how far your price can move before the contribution disappears.
Starter checklist
Today, no setup
- Write your price, the cost of one sale in parts, your monthly costs and your startup costs on one page
- Write the daily sales you expect in each of the first 12 months, and the most you can ship in a day
- List the objections you already hear from buyers
Set up
- Install Claude Code and run
claude --version - Open an empty project folder in Claude Code
- Install Founder Skill and type /founder to see the skills
Run
- Send the competitors prompt and open three links
- Send the CFO prompt and check contribution and break-even by hand
- Send the consumer prompt, read the profile first, and take the 20-buyer pass if offered
Decide
- Read the three files side by side
- Test the answer with real people before ordering stock
Troubleshooting
| What you see | What to do |
|---|---|
| /founder shows nothing | Restart Claude Code and run the two plugin lines again. If you copied folders, check they sit in ~/.claude/skills |
| claude is not found | The install folder is not on your PATH. Open a new terminal window, then check Anthropic's installation troubleshooting page |
| The competitors list is short or a site could not be reached | The skill reads public pages only and says so when a source fails. Add the rivals you already know and ask it to check them |
| The CFO skill asks for a number | It does not guess a number only you know. Give it the quote or invoice figure |
| The CFO skill shows a flag | Flags cover a price below cost, break-even above capacity, a first-year loss and payback beyond 12 months. Change the input behind it and run it again |
| A Python error appears | The tools need Python 3.8 or newer. Check with python3 --version |
| Claude Code asks permission for every file | Each buyer saves a file. In the terminal, press Shift+Tab to switch to accept-edits mode |
| The consumer run is slow or uses a lot of your plan | A full run is 100 helper calls. Ask for the 20-buyer quick pass first |
| Some buyers are missing from the results | A buyer that fails twice is left out and reported. Read that report before trusting the headline |
| Results change when you run it again | The buyers are simulated, so a new run can give a different headline. Compare runs on the shape of the result, not the last point |
Honest limits
- Free software is not free to run. The skills cost nothing, but a full consumer run and the competitor research use your Claude plan.
- The buyers are simulated. The README says the panel is not a forecast and that the buy rate should be read as an upper bound. Panel quotes are research, not testimonials, and should never be shown as customer reviews.
- Competitor research uses public pages only. Private, paywalled and login-only data is out of reach. It covers roughly 10 competitors, so a very crowded market is only partly covered, and prices are for the closest comparable item.
- The CFO maths is only as good as your numbers. Check costs with suppliers. Prices are before tax. Check payroll and tax with an accountant.
- It is not advice. The README says Founder Skill is not financial, legal or tax advice. Permits and trademarks depend on where you are.
- The consumer skill needs Claude Code. Pasting a single SKILL.md into a normal chat loses the helpers and the Python tools.
- The example is invented. Fernway, the harness, its costs and its sales plan were made up for the reel. The hand-worked figures in section 6 come from those made-up inputs.
- Founder Skill is one person's project. Commands and behaviour can change. Follow the repository where it differs from this guide.
How this guide was made. This guide is built from the Founder Skill README and the instruction files for the competitors, CFO and consumer skills on GitHub, plus Anthropic's Claude Code overview page. Sources were checked on November 7, 2026. No skill was run for this guide. The CFO arithmetic in section 6 was worked out separately from the reel's inputs.
Links
- Founder Skill on GitHub
- Competitors skill instructions
- CFO skill instructions
- Consumer skill instructions
- Claude Code overview and install (Anthropic)
- Claude Code installation troubleshooting (Anthropic)
What this changes
Most products learn what they needed after they are live. Three files, a competitor price range, a break-even number and a list of reasons buyers pass, give you something to decide on first.
Run the three checks on one product idea this week, and order stock only after the numbers and the buyers agree.



