Review product risk before AI agents build features

This skill helps an AI coding workflow pause before implementation to check whether a feature, product, or tool is actually worth building - focusing on product risk, meaning who needs it, what they use today, why they would switch, how distribution works, and what evidence would de-risk starting, rather than code structure. It runs a four-step process: identify the build bet by restating the product or feature in one concrete sentence naming the intended user, their job to be finished, and their current workaround or competitor; check the main risks across demand, workflow fit, willingness to switch, distribution, pricing, data access, and operational burden, preferring specific doubts over generic brainstorming; decide the smallest useful next validation step, such as a buyer conversation, landing page test, manual concierge workflow, prototype, waitlist, paid pilot, or narrow internal trial; and then decide whether to continue into implementation with assumptions written down, or recommend a smaller experiment first if risk is high or evidence is weak. Two worked examples show this applied to a SaaS dashboard feature request - checking which role needs it weekly, what they use today, what decision it changes, willingness to pay, and the smallest manual report proving repeat use - and an internal CRM request - what breaks in the current spreadsheet, daily user count, data sync needs, the process change required, and whether a no-code workflow can prove the need first.

Use it when a user asks an AI coding assistant to build a new app, feature, internal tool, SaaS, or side project, especially when the buyer, workflow, distribution path, or switching reason is still vague - always before writing code. Scale the review to project size: don't treat a small internal workflow like a startup pitch, and if the user already has strong evidence and a clear spec, keep the review short and move to implementation. It does not replace customer research, legal review, financial advice, or domain expert review, and cannot prove demand by itself - it only helps surface assumptions and choose a smaller validation step. Never fabricate market size, revenue, competitor traction, or buyer quotes to make an idea look validated.

Given a build request, it produces a one-sentence restated build bet, a risk assessment across the seven risk dimensions, a recommended smallest next validation step, and a continue-or-stop recommendation with assumptions documented.

Ships as a standalone skill repository with an npx installer for several coding assistants; safe to run as a planning-only layer since it requires no credentials, network access, or file mutation. Related skills include a SaaS MVP launcher for moving from validation into MVP planning, and a UX research methodology skill for structured user research when that's the next step.

Teams and solo builders using an AI coding assistant who want a forced pause to challenge product assumptions - user, job, alternative, switching reason - before committing engineering time, and to pick the riskiest assumption to test first rather than boiling the ocean.