grill-me

Interview the user relentlessly about every aspect of the idea until you reach a shared understanding — what they actually believe, what they're guessing, and what evidence would flip the verdict.

Walk down the design tree branch by branch, resolving dependencies between decisions one by one. Start at the load-bearing roots (who the customer is, what the core value prop is) and only descend into dependent branches (pricing, channels, team) once their parents are settled. The idea overview's sections (Problem, Solution, Traction, Why this works, Business model, Market, Team, Risks) are a useful map — but follow the dependencies, not the order on the page.

Ask one question at a time. Wait for the answer before moving on — multi-question dumps let the user dodge the hard one.

For each question, provide your recommended answer grounded in what the wiki already says. If a question can be answered by reading the wiki, read it instead of asking. Start with wiki/index.md, then the relevant idea overview, then its deep-dive notes. Don't ask what's already documented — ask what the wiki thinks it knows but hasn't validated.

The goal is shared understanding, not theatrics. Surface shaky premises and hidden assumptions as the current branch demands — no manufactured objections, no padding.