Find a buyer question, recover the baseline from your notes, or check the price of a pilot. Each guide ends with something you can verify before acting.
How do MCP servers and skills work?
Connect an MCP server. Model Context Protocol (MCP) lets an assistant call tools in another service. The listings here were chosen for publisher documentation and a first task with a checkable result. Open the publisher guide, connect in a compatible assistant, and try the request with your own account.
Use a skill. Download its ZIP. In a client that accepts skill ZIPs, import it using the client’s skills settings. For a folder-based client, extract the named folder into its documented skills directory. In an ordinary chat, paste the instructions from SKILL.md, then the sample request.
See how makers describe competing offers and what prospective buyers still ask.
How to use it
What you need
Product Hunt account and a client that supports remote MCP with OAuth.
Server URL
https://mcp.producthunt.com/mcp
Set up and try it
Add the server in your assistant’s MCP settings. Choose OAuth and approve access on Product Hunt.
Start a new conversation with the connection enabled. Run the first request below.
First request
Compare three Product Hunt launches for independent workshop organizers. List the promised result, one audience question if present and the source. End with an interview question I can test.
What you should get
A competitor table and one question for a real buyer conversation.
Check the result
Treat comments and votes as research leads. Record purchase evidence separately.
Catch a price or billing interval mismatch before inviting buyers to a pilot.
How to use it
What you need
A Stripe sandbox, an existing test product and permission to view prices.
Server URL
https://mcp.stripe.com
Set up and try it
Add Stripe’s remote server in a supported client. Sign in with OAuth, select the sandbox and review the permissions shown before connecting.
Start a new conversation with the connection enabled. Run the first request below.
First request
In my sandbox, list the active prices for my workshop product. Include amount, currency and one-time or recurring billing. Do not create or change anything.
What you should get
A price inventory you can compare with the offer copy.
Check the result
Match the product, currency and billing interval to the offer and confirm the records are in test mode.
Find specific places where your intended buyer already asks about the problem.
How to use it
What you need
A remote MCP client. Exa offers a free starting tier with usage limits.
Server URL
https://mcp.exa.ai/mcp
Set up and try it
Add the HTTPS server to your client and confirm that web search is available. Start with a single, narrow query.
Start a new conversation with the connection enabled. Run the first request below.
First request
Find three public communities for independent workshop hosts. Link to their participation guidelines and identify where questions about filling seats appear. Flag unclear posting rules.
What you should get
A shortlist of channels, their rules and evidence of audience fit.
Check the result
Open the guidelines before choosing a channel. Leave a missing rule unresolved.
Turn a batch of visits, purchases and costs into a decision you can revisit with the original evidence.
How to use it
What you need
Hypothesis, start and end dates, audience, unit of exposure, success rule agreed before launch, observed counts, costs and source for each number.
How the skill works
Write the hypothesis as a prediction for one audience and one time window. Copy the original success rule verbatim. If the team never set a rule, label the run exploratory.
Put the observations in a table with metric, count, time window and source. Use unique eligible visitors for a visitor conversion rate. Flag duplicate orders, refunds, missing costs and overlapping dates.
Calculate the primary rate from counts and show the arithmetic. With a zero or unknown denominator, report the rate as unavailable. Calculate revenue only from observed paid orders and actual prices.
Apply the original rule and name the strongest alternative explanation. Return a decision, its evidence and one next test with a measurable stopping condition.
Try the skill
Score my sample seven-day offer test. Hypothesis: at least 4% of unique visitors buy the $15 checklist. Continue only with at least 100 unique visitors and a 4% purchase rate. Analytics shows 120 unique visitors. Checkout shows four paid orders at $15, zero refunds. Ad spend was $18. Fees are missing.
Worked example · sample data
Stop this version of the offer. Four purchases from 120 visitors is 3.33%, below the 4% rule. Gross sales were $60. Sales less ad spend were $42 before fees and other costs. Traffic sources were not provided. Check whether one source dominates before attributing the result to the offer.
Check the result
The decision cites the original rule, shows a count and denominator, and identifies missing cost data before reporting a profit figure.
Trace each product assumption back to a buyer's words or behavior and identify the question worth asking next.
How to use it
What you need
Anonymized notes with participant IDs, verbatim quotes where available, interviewer summaries, buyer segment and the product assumption under review.
How the skill works
Remove direct identifiers. Assign one ID per participant and keep direct quotations separate from interviewer paraphrases.
Extract the last concrete incident, current workaround, consequence and any observed spending or time. Mark a missing field unknown. Treat a compliment or hypothetical intention as a separate type of evidence.
Group recurring problems and count distinct participants, including anyone who contradicted the pattern. Return a table with problem, participant IDs, evidence type and counterexample.
Pick one assumption to test. Write a neutral question about the participant's last real experience and name the observable behavior that would strengthen or weaken the assumption.
Try the skill
Assess a reminder kit from these sample interviews. P1 said 'I called six people the night before.' P2 said 'Our booking app reminds everyone.' P3 said 'I copied last month’s text into eight separate messages.' No prices or purchase commitments were discussed.
Worked example · sample data
Manual reminder work appeared for P1 and P3, two of three participants. P2 already has an automated workaround. Time spent and willingness to pay are unknown. Follow-up: 'Walk me through the last set of reminders you sent, including when you started and finished.'
Check the result
Every finding names the supporting participant IDs, repeated mentions by the same person count once, and no price commitment appears without evidence.
Make a channel decision using comparable visits and outcomes with an explicit time or spending limit.
How to use it
What you need
Two eligible channels, posting rules, offer and landing page, measurable conversion, tracking method, available hours, budget and test window.
How the skill works
Check that the proposed placement is allowed in each channel. Hold the core offer and landing page constant, and adapt the opening line to each audience.
Assign each channel a unique tracking link or code. Define the denominator, conversion event, minimum sample, stop date and total budget before drafting posts.
Create a two-row launch sheet with placement, copy, tracking code, owner and timing. Draft only. If permission or attribution is unclear, flag the dependency.
At review, compare eligible clicks, attributable orders, spend and effort. Show raw counts beside rates. When tracking fails or either sample misses the gate, report the result as inconclusive.
Try the skill
Plan a seven-day test for a $19 worksheet through my opt-in newsletter and my own LinkedIn feed. Same landing page, two hours total, no ad spend. Compare purchases per unique landing visitor. Require 50 unique visitors per channel. Example results: email 3/60, LinkedIn 1/20.
Worked example · sample data
Use utm_source=newsletter and utm_source=linkedin with utm_campaign=worksheet_test. Email recorded 5.0% from 60 visitors. LinkedIn recorded 5.0% from 20. The second channel missed the 50-visitor gate, so the comparison is inconclusive.
Check the result
Both channels use the same conversion and attribution window. The decision respects each minimum sample, and every piece of draft copy has an identifiable destination.
No matches. Try a broader task or clear the filters.
Free download · The Experiment Scorecard
Know if your side idea is actually working.
A 6-page scorecard for testing one income idea: the signal ladder, a four-week evidence log, the true-cost check, and a clear continue / change / pause decision. Download it now, with no signup required. Follow Ashley’s experiments in the journal and free RSS feed.