brainstorm

name	brainstorm
description	Research-driven brainstorming for feature design. Use when designing a new feature, exploring approaches, or making design decisions. Automatically researches competitors before asking design questions. Triggers on "brainstorm", "design a feature", "how should we build X", or any request to explore approaches for a feature.

Brainstorm: Research-Driven Design

Transform feature ideas into well-researched designs. Unlike traditional brainstorming that asks endless questions, this skill researches competitors first so design decisions are informed by industry patterns.

Announce at start: "I'm using the brainstorm skill to design this with research."

The Problem This Solves

Traditional brainstorming:

Claude: "How should we handle partial shipments?"
User:  "Good question, how does SAP do it?"
Claude: [researches]
Claude: "What about backorders?"
User:  "How does NetSuite handle that?"
Claude: [researches again]
... repeat 10 times ...

This skill front-loads that research.

Workflow

┌──────────────┐    ┌──────────────┐    ┌──────────────┐    ┌──────────────┐
│ 1. Scope     │───▶│ 2. Research  │───▶│ 3. Design    │───▶│ 4. Document  │
│    Feature   │    │ Competitors  │    │ With Data    │    │ & Approve    │
└──────────────┘    └──────────────┘    └──────────────┘    └──────────────┘

Step 1: Scope the Feature

1.1 Understand the Request

Ask focused questions (max 3-4) to understand:

• What — The core capability
• Why — The problem it solves
• Where — Which part of Carbon (ERP, MES, Academy)
• Who — Primary users

1.2 Classify the Domain

Map to research domains:

Domain	Keywords	Competitors
Accounting	GL, AP, AR, invoicing, payments	SAP, NetSuite
Manufacturing	Jobs, work orders, BOMs, routing	SAP, Epicor
MES	Real-time tracking, scheduling, dispatch	SAP, Manufacturo, First Resonance
CNC/Sheet Metal	Quoting, estimating, nesting	SAP, Fulcrum, Paperless Parts
Quality	Inspection, SPC, FAI, GD&T	SAP, 1factory, HighQA
Inventory	Valuation, lots, warehousing	SAP, NetSuite, Fishbowl
Sales	Quotes, orders, pricing	SAP, NetSuite
Purchasing	POs, suppliers, receiving	SAP, NetSuite, Coupa

1.3 Identify Research Questions

Before researching, list the key design questions:

## Research Questions for [Feature]
1. [Data model question]
2. [Workflow question]
3. [Edge case question]
4. [Integration question]

Step 2: Research Competitors

2.1 Execute Research

Invoke /research [feature] targeting the identified questions.

Search patterns per competitor:

• [Competitor] [feature] documentation
• [Competitor] [feature] workflow
• [Competitor] [feature] best practices

2.2 Synthesize

Structure findings as:

1. Key Consensus — What all competitors agree on
2. Unique Approaches — Interesting variations
3. Answers to Questions — Direct answers from Step 1.3

Save to llm/research/[feature-slug].md.

Step 3: Design With Data

Now brainstorm with research in hand.

3.1 For Each Design Decision

1. State the question
2. Reference the research — "SAP and NetSuite both..."
3. Propose 2-3 approaches with trade-offs
4. Recommend one based on industry patterns

3.2 Cover These Areas

• Data Model — Tables, relationships, key fields
• Workflows — User journeys, state transitions
• Edge Cases — How competitors handle the tricky stuff
• Integration — How this connects to existing Carbon features

Step 4: Document & Approve

4.1 Write Design Spec

Save to docs/specs/[feature-name]-design.md:

# [Feature] Design Specification

## Summary
[One paragraph]

## Research Summary
[Key findings, link to full research]

## Design Decisions

### [Decision 1]
**Question:** [The design question]
**Industry Pattern:** [What competitors do]
**Our Approach:** [What we'll do and why]

## Data Model
[Schema]

## Workflows
[User journeys]

## Edge Cases
[Tricky stuff, informed by competitors]

4.2 Self-Review

Before presenting to user:

• No placeholders or TBDs
• All research questions answered
• Data model complete
• Edge cases addressed

4.3 Get Approval

Present the spec and wait for explicit approval before proceeding.

Output

Artifact	Location
Competitor research	llm/research/[feature-slug].md
Design specification	docs/specs/[feature-name]-design.md

Next Step

After approval, hand off to /plan to create the implementation plan.

When to Stop and Ask

• Feature is more complex than expected
• Decisions require business input (not just technical)
• Research reveals conflicting patterns with no clear winner
• Uncertain about something that affects other Carbon modules