mom-test
Talk to customers without leading them using Mom Test rules: discuss their life not your idea, ask about specifics in the past, and talk less
Install / Use
npx skills add wondelai/skills --skill mom-testInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
Education & ResearchSupported Platforms
Our assessment of mom-test
mom-test scores 87/100 on our quality scale, 243rd of 421 Education & Research skills we index.
Its SKILL.md is 16 KB long, well organised into 14 sections and no code examples: a thorough specification that gives an agent plenty to work with.
With 2,248 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so mom-test is actively maintained.
- Our last check on 2026-11-02 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-30. Automated pattern scan on 2026-09-29. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
mom-test compared with similar skills
All 4 of these similar skills score higher than mom-test; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| mom-test (this skill)by wondelai | 87 | 2.2k | 2mo ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 116.8k | 8d ago | CLAUDE.md |
| last30days-skillby mvanhorn | 100 | 68.0k | today | CLAUDE.md |
| Claude-Code-Game-Studiosby Donchitos | 100 | 39.7k | 7d ago | CLAUDE.md |
| x402by browser-use | 100 | 121.0k | 3d ago | SKILL.md |
Frequently asked questions
- How do I install mom-test?
- Run
npx skills add wondelai/skills --skill mom-test. The install tabs above show the steps for each supported agent. - Which AI agents does mom-test work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is mom-test safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is mom-test still maintained?
- The repository was last updated about 2 months ago, so mom-test is actively maintained.
Skill content
View source on GitHubname: mom-test description: 'Talk to customers without leading them using Mom Test rules: discuss their life not your idea, ask about specifics in the past, and talk less. Use when the user mentions "customer interviews", "validate my idea", "users say they want it but dont buy", "leading questions", "The Mom Test", "customer feedback bias", or "interview script". Also trigger when preparing user-research questions, interpreting ambiguous feedback, or designing customer-discovery that avoids false positives. Covers commitment and advancement, avoiding compliments, and extracting signal from noise. For product-market fit, see jobs-to-be-done. For rapid prototype testing, see design-sprint.' license: MIT metadata: author: wondelai version: "1.4.0"
The Mom Test Framework
Framework for customer conversations that won't lead you astray, based on a fundamental truth: everyone is lying to you -- not maliciously, but because you're asking the wrong questions. The Mom Test provides rules for asking questions so good that even your mom can't lie to you.
Core Principle
Good customer conversations are about their life, not your idea. The moment you mention what you're building, people switch from sharing truth to performing politeness. Talk about their problems, their lives, and their existing behavior instead of pitching, and ask about specifics in the past, not hypotheticals about the future. Above all, talk less and listen more.
Scoring
Goal: 7/7. Score a conversation (or interview plan) by the seven-row Quick Diagnostic below: 1 point per row that passes.
- 6-7 = focused on their life and past behavior, concrete facts captured, a real commitment (time/reputation/money) secured, they talked 80%+, and beliefs got updated.
- 4-5 = some past-behavior facts but leaking into hypotheticals, compliments accepted as signal, or ending without an ask.
- <=3 = a pitch in disguise: leading questions, opinions and fluff, a polite zombie lead, no learning.
Always state the current score out of 7, name the failing diagnostic rows, and give the specific fix for each.
Framework Sections
1. The Mom Test Rules
Core concept: Three rules that make it impossible for even your most supportive loved ones to give you false validation, shifting conversations from opinion-gathering to fact-finding.
Why it works: People are unreliable predictors of their own future behavior, so opinions are worthless. Past behavior is the only reliable data and can genuinely inform product decisions.
Key insights:
- Rule 1: Talk about their life, not your idea -- never mention your solution until the end, if at all
- Rule 2: Ask about specifics in the past, not generics or hypotheticals about the future
- Rule 3: Talk less, listen more -- aim for them to speak 80% of the time
- A question fails the Mom Test if the answer is always "yes" regardless of whether the business will succeed
- Good questions could potentially destroy your currently imagined business
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Idea validation | Ask about the problem, never the solution | "Tell me about the last time you tried to [problem area]" not "Would you use an app that does X?" | | Feature prioritization | Discover what people do vs. what they say | "Walk me through how you handled this last week" | | Pricing research | Anchor to existing spending behavior | "What are you currently paying to solve this?" not "Would you pay $X?" |
Copy patterns:
- "Tell me about the last time you..."
- "What else have you tried?"
- "Why does that bother you?"
See: references/question-patterns.md when drafting an interview script -- a 5-tier question hierarchy, domain-specific question banks (SaaS/consumer/marketplace), and four formulation exercises.
2. Good vs Bad Questions
Core concept: Most interview questions are broken because they ask people to predict the future, evaluate hypothetical products, or confirm your assumptions. Good questions anchor in observable past behavior and extract concrete facts.
Why it works: Asking "would you buy this?" is like asking "will you go to the gym next week?" -- the answer is always yes, the follow-through rarely there. Behavior that already happened can't be rationalized away.
Key insights:
- Bad: "Do you think it's a good idea?" -- always gets a yes
- Bad: "Would you buy a product that does X?" / "How much would you pay?" -- hypothetical, anchored to please you
- Good: "How are you dealing with this problem today?" -- reveals actual behavior
- Good: "What have you tried before and why did you stop?" -- reveals past decisions
- Good: "Where does the money come from for solutions like this?" -- reveals real budgets
- The scariest questions -- ones with the power to change what you're building -- produce the most useful data
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Problem validation | Confirm the problem exists and matters | "When did this last come up? What did you do? What didn't work?" | | Market sizing | Check if enough people share the problem | "Who else in your industry deals with this? How do they handle it?" | | Competitive analysis | Find real alternatives already in use | "What tools/processes do you currently use for this?" |
Copy patterns:
- "What's the hardest part about [doing this thing]?"
- "How often does this come up?"
- "Walk me through what happened the last time this came up"
3. Avoiding Compliments and Opinions
Core concept: Three types of bad data feel like progress but actively mislead: compliments ("That's a great idea!"), fluff (hypotheticals, maybes, future promises), and ideas (feature requests disconnected from real problems). Deflecting these and digging for truth is the core skill.
Why it works: Compliments are the fool's gold of customer development -- they feel amazing but contain zero information about whether anyone will pay or use the product. Only specifics about real past behavior and genuine commitments provide signal.
Key insights:
- Compliments: deflect immediately and return to concrete facts about how they handle the problem today
- Fluff: generic claims ("I usually," "I always," "I would never") are worthless without a specific instance
- Ideas: dig into the motivation behind every feature request -- what's driving it, when they last needed it
- Fishing for compliments ("Don't you think this would be useful?") is unconscious validation-seeking
- Symptom of a bad conversation: you walk away feeling great but with no concrete facts or commitments
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Post-demo feedback | Deflect "this looks awesome" | "Thanks! What part of your current workflow would this replace?" | | Feature requests | Dig for the underlying job | "Why do you want that? Can you show me the last time you needed it?" | | Investor conversations | Separate encouragement from interest | Ask for customer intros, not "great idea" feedback |
Copy patterns:
- "Thanks, but to make sure I'm not wasting your time -- what does your current process look like?"
- "When you say you'd 'definitely' use this, what would you stop using?"
- "That's a great feature idea -- what problem would it solve for you specifically?"
See: references/avoiding-bad-data.md when a conversation feels good but yields no facts -- how to spot and deflect the three bad-data types (compliments, fluff, ideas) in real time.
4. Commitment and Advancement
Core concept: The currency of a customer conversation is commitment, not compliments. End every conversation with a clear advance toward adoption or a clear rejection -- the worst outcome is a "zombie lead" who is polite but never commits.
Why it works: Saying "I'd definitely buy that" costs nothing; offering an intro, a deposit, or a pilot invests something real. Commitment closes the dangerous gap between what people say and what they do.
Key insights:
- Commitment currencies: time (meeting, trial), reputation (intro, testimonial), money (deposit, pre-order, letter of intent)
- Advancing moves the relationship toward a sale; spinning wheels produces pleasant, useless meetings
- Know your "ask" before the meeting -- the minimum commitment that proves this is real
- A "no" is more valuable than a "maybe" -- you can learn from it and move on
- If they won't give you their time, they definitely won't give you their money
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Early validation | Request a commitment that tests interest | "Can I follow up with a prototype next week for 15 minutes of your time?" | | B2B sales | Advance toward the decision-maker | "Could you introduce me to the person who handles the budget for this?" | | Pre-launch | Collect pre-orders or letters of intent | "Launching in 8 weeks -- want to join the first cohort at 40% off?" |
Copy patterns:
- "Who else should I talk to about this?"
- "Would you be willing to try a prototype next week?"
- "If I built this, would you be willing to pilot it for 30 days?"
See: references/commitment-advancement.md when a conversation ends without a commitment -- the currency ladder (time/reputation/money) and scripts for advancing instead of spinning wheels.
5. Finding Conversations
Core concept: The best customer conversations happen casually -- warm intros, industry events, online communities, coffee. Formal "customer interview" framing triggers performance mode; casual framing produces honest data.
Why it works: "Can I interview you about your problems?" makes people polished and guarded; "I'm trying to learn about the industry -- can I buy you coffee?" makes them open up. The framing determines the quality of the data.
Key insights:
- Cold outreach: keep it short, lead with their expertise, don't pitch
- Warm intros are the best source -- one well-connected advisor can open dozens of doors
- Go where customers already gather: industry events, meetups, online communities (participate genuinely first)
- "I'm trying to learn" beats "I'm doing customer research"
- Use the five-part structure for getting meetings: vision / framing / weakness / pedestal / ask
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Pre-idea exploration | Immerse in the target community | 3 industry events and 20 casual conversations before writing code | | B2B prospecting | Warm intros through advisors | "Our advisor [Name] suggested I ask how you handle [problem area]" | | Consumer research | Intercept at the point of behavior | Talk to people in line at the store, the gym, the coworking space |
Copy patterns:
- "I'm researching how [industry] handles [problem] -- could I learn from your experience over a 15-minute coffee?"
- "[Mutual contact] suggested I talk to you because you know a lot about [area]"
- "I'm not trying to sell anything -- I'm just trying to understand the space"
Ethical boundary: Never disguise a sales call as a learning conversation -- if you already have a product and are selling, be transparent.
See: references/finding-conversations.md when you need to source interviews -- cold vs warm outreach templates, the five-part meeting ask (vision/framing/weakness/pedestal/ask), and how to keep it casual.
6. Processing and Learning
Core concept: Conversations are only useful if processed: distill raw notes into beliefs, update them regularly, and share with your team. Without a system you'll cherry-pick quotes that confirm your biases.
Why it works: Memory is biased toward recent and emotionally charged
Truncated for display — read the full file on GitHub.
Related Skills
taste-skill
116.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
last30days-skill
68.0kAI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary
Claude-Code-Game-Studios
39.7kTurn Claude Code into a full game dev studio — 49 AI agents, 72 workflow skills, and a complete coordination system mirroring real studio hierarchy.
x402
121.0kSet up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
