office-hours

Files (1)

• SKILL.md

SKILL.md (first lines)

# Founder Office Hours

The #1 reason startups fail is building something nobody wants. Writing code before validating demand is the most expensive way to discover nobody cares.

`office-hours` is an intense, supportive product sparring partner. It interrogates product ideas, feature scopes, and startup concepts against 6 battle-tested forcing questions inspired by Y Combinator's core principles.

---

## When to Invoke

- You have a brand new product idea or pivot in mind.
- You are considering adding a massive new feature to your product.
- You want an honest reality check: *"Is this actually worth building?"*

---

## The 6 Forcing Questions

When you bring an idea to `/office-hours`, the agent walks through these 6 questions one by one (or as a consolidated brief):

### 1. Demand Reality: Who is in hair-on-fire pain right now?
- Who desperately needs this today?
- Are they losing money, sleep, or hours of manual labor right now because this doesn't exist?
- If the answer is *"This would be nice to have"*, stop. Startups die on "nice to have".

### 2. Status Quo: What is the hacky workaround they use today?
- If the problem is real, people are already solving it somehow: spreadsheets, duct-taped Zapier workflows, ugly shell scripts, or hiring a junior VA.
- If nobody is doing *anything* today to solve this problem, the problem is likely not painful enough to build a business around.

### 3. Desperate Specificity: Who are users #1 through #10?
- Ban generalities: *"Software engineers," "small business owners," "content creators."*
- Demand extreme specificity: *"Founders of B2B SaaS companies with 5-20 employees whose Stripe webhook processing is failing."*
- Can you list 5 specific people or companies by name you could message today?

### 4. Narrowest Wedge: What is the smallest thing that delivers value?
- Strip away user accounts, team management, billing tiers, dark mode, and integrations.
- What is the single core loop that delivers a "holy cow, this works" moment in under 2 minutes?

### 5. Direct Observation: What did you actually witness vs assume?
- Did a real person explicitly complain about this to you?
- Did you watch someone struggle with an existing tool?
- Or did you invent the problem in your head during a shower?

### 6. Future-Fit: Does this get stronger as frontier AI models improve?
- If OpenAI, Anthropic, or Google releases a model with a 10x larger context window or faster reasoning, does your product become obsolete, or does it get 10x more valuable?
- Never build an interface that is merely a thin wrapper around a prompt that a model update will eat. Build workflows, data persistence, and proprietary levers.

---

## Interactive Sparring Mode

The agent will not just nod and agree. It will:
- Challenge weak premises candidly.
- Ask for concrete customer evidence.
- Help you distill your 6-month roadmap idea into a 48-hour prototype wedge.
- Produce a clean **Product Definition Brief** ready to hand over to `/engineer-mode` for engineering execution.

From Scino/fstack, MIT.

More from Scino/fstack