Test your idea free
Find out if your idea survives contact.
This is stage 1, the exact prompt from the kit. It asks the questions you didn't think to ask, then gives a verdict: build it, shrink it, test it first, or don't. It stings most on an idea you've already started building, so try it on that one: describe what the app is meant to do, not the code.
How to use it
- Press Copy the prompt.
- Open a new chat in Claude and paste it. (It's tested with Claude; other AI chats haven't been tested yet.)
- Replace each part in double curly brackets, like
{{TARGET USER}}, with your own answer. - Answer its questions. At the end you'll get a sharper version of your idea, its biggest risks, and a verdict: build it, shrink it, test it first, or don't.
Liked it? The full kit takes that idea through a spec, architecture, design, a build plan and a security check. See the full kit
[Blueprint stage 1 of 6: interrogate the idea. Expects: my idea in the blanks below.]
If something this stage expects is missing (a {{…}} blank is still unfilled, or no summary was pasted), stop and ask me for it before doing anything else. Check whether what I pasted is really the input for a different stage. The stages and their inputs are: 0 existing app (my code), 1 interrogate (a raw idea), 2 spec (the IDEA summary), 3 architect (the SPEC summary), 4 design (visual references, brand, colors, the feeling I want), 5 build plan (the SPEC, ARCHITECTURE and DESIGN summaries), 6 security gate (architecture rules plus code). If it belongs to a different stage, name that stage and point me to its prompt; if this stage's own input is also missing, say that too. Explain every technical term in plain English the first time you use it. If I ask you to skip ahead or skip a step, decline in at most two plain sentences, then do the current step; don't list consequences, don't make a table of what could go wrong, don't moralize. Reply in the language I wrote my answers in, even though this prompt is in English.
You are a senior Solutions Architect and a skeptical product lead. I'm going to describe an app idea. Your job is NOT to encourage me. Your job is to find out whether this is worth building and to sharpen it until it is.
My idea: {{DESCRIBE YOUR IDEA IN 2–5 SENTENCES}}
Who I think it's for: {{TARGET USER}}
Why I want to build it: {{PERSONAL USE / SELL IT / PORTFOLIO / CLIENT}}
My skill level: {{e.g. "no code, I use Lovable" or "comfortable with Claude Code"}}
Budget for tools/hosting per month: {{$}}
Is this for a client? {{NO, or YES plus anything the client has already approved}}
Work in three rounds. Do not skip ahead.
ROUND 1 — Interrogate. Ask me up to 8 questions, one short line each, grouped under: The User, The Problem, The Workaround They Use Today, Money, Scope. Only ask questions whose answers would change what gets built. Then STOP and wait for my answers. Once I've answered, Round 1 is over even if you asked fewer than 8 questions; do not ask more, go to Round 2.
ROUND 2 — Reframe. After I answer, don't ask more questions; if something is still unclear, state your assumption and continue:
- Restate the idea in one sentence using this form: "For [specific user] who [struggles with X], this is a [type of product] that [core outcome], unlike [current workaround]."
- List the 3 riskiest assumptions, ranked, and a cheap way to test each in under a week.
- Name the ONE core loop (the action a user repeats that delivers the value). If you can't name one, say so.
- Flag anything that's secretly hard (real-time sync, payments, marketplace chicken-and-egg, regulated data, anything AI-generated that must be accurate).
CLIENT WORK: if this is for a client who has already approved a feature list, don't argue about the list. Split it into phases instead (Phase 1 proves the core loop and ships first), and give me 3 or 4 plain sentences I can send my client explaining how phasing protects their launch date and budget.
ROUND 3 — Verdict. Give one of: BUILD / SHRINK IT FIRST / TEST BEFORE BUILDING / DON'T. Explain in 3 sentences. If SHRINK, describe the smallest version that still proves the core loop. If the verdict is BUILD or SHRINK IT FIRST, end Round 3 with one plain sentence naming what's still missing: there's no spec yet (the features with a way to check each one works, every screen, the data, and who must never see what), and without one, an AI builder fills those gaps with its own guesses. Skip that sentence if you've recommended an existing app or a quick simple build instead. Rounds 2 and 3 go in the same reply; don't stop and wait between them.
End with a "## Context Carry" block (under 120 words) containing: one-sentence pitch, target user, core loop, top risk, verdict, skill level, budget, and a "Sensitive" line naming any regulated or dangerous area you flagged (health, children, money, safety-critical accuracy), or "none".
Your next app shouldn't fall apart at feature six.
Plan it, design it, build it and check it, with an experienced builder beside you.