---
name: 1mancompany/project-brainstorming-requirements-discovery-before-execution
source: https://app.decimal.ai/s/1mancompany-project-brainstorming-requirements-discovery-before-execution@1/SKILL.md
source_sha256: f97f94309c8e
---

# Project Brainstorming — Requirements Discovery Before Execution

Turn vague ideas into clear specs with acceptance criteria through collaborative
dialogue with the CEO. Understand what to build before building it.

<HARD-GATE>
Do NOT dispatch any project work to the team until the CEO has approved your
proposed plan. This applies to EVERY project regardless of perceived simplicity.
"Simple" projects are where unexamined assumptions cause the most wasted work.
The design can be SHORT (a few sentences for truly simple projects), but you
MUST present it and get CEO approval before dispatching.
</HARD-GATE>

## When to Use This Skill

**USE brainstorming for:**
- New product/feature requests ("build X", "create Y", "design Z")
- Multi-step projects that involve multiple team members
- Ambiguous tasks where CEO intent needs clarification

**SKIP brainstorming for:**
- Simple operational tasks ("check status", "send email", "look up info")
- Urgent fixes ("the site is down", "fix this bug now")
- Tasks where CEO gave explicit, detailed instructions with clear criteria
- Follow-up tasks on existing projects where scope is already defined
- CEO explicitly says "just do it" or "skip the questions"

---

## Phase 1: Understand Context (no CEO interaction)

Before asking the CEO anything:
1. Read existing project files if this is a follow-up
2. Check what team members are available (use list_colleagues)
3. Form your own understanding of the task

---

## Phase 2: Ask Clarifying Questions (ONE AT A TIME)

Ask the CEO questions via `dispatch_child(target_employee_id="00001", ...)`.
The system preserves your conversation history — when CEO replies, you will
see ALL prior CEO replies in your task context under [CEO REPLY] sections.

### Rules:
- **One question per dispatch.** Do NOT bundle multiple questions.
- **Prefer multiple choice** when possible — easier for CEO to answer.
- **Max 3-4 questions** — respect CEO's time. Stop when you have enough context.
- **Check your [CEO REPLY] sections** — do NOT re-ask questions CEO already answered.

### What to ask about:
1. **Purpose & Users**: Who is this for? What problem does it solve?
2. **Scope**: What's the minimum viable version? What can wait for later?
3. **Constraints**: Tech stack, timeline, budget, hosting preferences?
4. **Success criteria**: How will the CEO know this is done and successful?
5. **Priority**: Speed vs. quality? MVP vs. polished?

### Good question examples:
- "What's the primary goal? (A) Generate revenue (B) Attract users (C) Internal tool (D) Research/learning"
- "Who is the target user? (A) End consumers (B) Businesses (C) Internal team (D) Developers"
- "What's the priority? (A) Ship fast with basic quality (B) Take time for polish"

---

## Phase 3: Challenge Premises

Before proposing solutions, question your own assumptions:

- **Is the stated problem the real problem?**
- **Does this need to be built at all?** Is there a simpler approach?
- **Is the scope right?** If too large, propose breaking into phases.

If you identify a premise worth challenging, raise it with the CEO via
dispatch_child(target_employee_id="00001").

---

## Phase 4: Propose 2-3 Approaches (MANDATORY)

You MUST present at least 2 approaches with trade-offs:

```
dispatch_child(
    target_employee_id="00001",
    description="""
Based on our discussion, here are the approaches I see:

## Approach A: [Name] (Recommended)
**What**: [2-3 sentences]
**Pros**: [bullet points]
**Cons**: [bullet points]
**Team**: [who does what]

## Approach B: [Name]
**What**: [2-3 sentences]
**Pros**: [bullet points]
**Cons**: [bullet points]
**Team**: [who does what]

**My recommendation**: Approach [X] because [one-line reason].

**Proposed Acceptance Criteria**:
1. [measurable, verifiable criterion]
2. [measurable, verifiable criterion]
3. [measurable, verifiable criterion]

Please choose an approach and confirm/adjust the acceptance criteria.
""",
    acceptance_criteria=["CEO selects approach and approves acceptance criteria"]
)
```

---

## Phase 5: Execute After CEO Confirms

After CEO approves:
1. Call `set_project_name()` with a concise 2-6 word name
2. Incorporate CEO feedback into the final plan
3. Dispatch to O-level executives with the confirmed acceptance criteria
4. Save a brief `design.md` to the project workspace for medium/large projects

---

## Key Principles

- **CEO's time is the scarcest resource.** Keep questions focused and minimal.
- **Never assume scope.** "Build a website" could mean a landing page or a full platform.
- **Always present 2+ approaches** with trade-offs.
- **Acceptance criteria are a contract.** Once CEO confirms, deliver against them.
- **Check [CEO REPLY] sections** before each turn — don't lose context or re-ask.
- **If CEO says "just do it"** — respect that and dispatch immediately.
- **Scope decomposition.** If too large, propose phases. Each phase gets its own cycle.