Startup Design Skill

An AI agent skill for comprehensive startup design and idea validation. Takes a raw idea from concept to validated business plan through structured research, strategic frameworks, and honest assessment.

Built for founders, solopreneurs, and product people who want to validate ideas rigorously before investing time and money. Works with Claude Code and any agent that supports skills.

Contributions welcome! Found a way to improve the skill or have a new idea? Open a PR.

Run into a problem or have a question? Open an issue.

What It Does

The skill guides an AI agent through 8 structured phases, producing a complete set of markdown documents:

INTAKE -> BRAINSTORM -> RESEARCH -> STRATEGY -> BRAND -> PRODUCT -> FINANCIAL -> VALIDATION

Each phase builds on the previous one. Sessions can be interrupted and resumed from checkpoints.

Phases

Phase	What It Produces	Key Output Files
1. Intake Interview	Deep understanding of the idea, founder, and constraints	00-intake/brief.md
2. Brainstorm	5-8 variations of the idea before committing	00-intake/brainstorm.md
3. Market Research	Multi-agent web research in 4 waves (11 agents)	01-discovery/*.md
4. Strategy	Lean Canvas, positioning, business model, GTM	02-strategy/*.md
5. Brand	Mission, vision, tone of voice, personality	03-brand/*.md
6. Product	MVP definition, feature prioritization, user journey	04-product/*.md
7. Financial	Revenue model, cost structure, projections	05-financial/*.md
8. Validation	Experiments, risk analysis, scorecard	06-validation/*.md

Modes

• Full Mode (default): All 8 phases, thorough analysis
• Fast Track: Compressed version for quick go/no-go decisions

Features

• Multi-language support (auto-detects from user input, defaults to English)
• Progress tracking via PROGRESS.md for session resumption
• Multi-agent parallel web research with source quality tiers
• Strategic frameworks: Lean Canvas, April Dunford Positioning, Value Proposition Canvas, RICE/MoSCoW
• Honest assessment with risk analysis and validation experiments

Installation

Option 1: CLI Install (Recommended)

Use npx skills to install skills directly:

# Install the skill
npx skills add ferdinandobons/startup-skill

# List installed skills
npx skills list

This automatically installs to your .agents/skills/ directory (and symlinks into .claude/skills/ for Claude Code compatibility).

Option 2: Claude Code Plugin

Install via Claude Code's built-in plugin system:

# From the Claude Code CLI
claude plugin marketplace add ferdinandobons/startup-skill
claude plugin install startup-design

Option 3: Clone and Copy

Clone the entire repo and copy the skill folder:

git clone https://github.com/ferdinandobons/startup-skill.git
cp -r startup-skill/startup-design .agents/skills/

Option 4: Git Submodule

Add as a submodule for easy updates:

git submodule add https://github.com/ferdinandobons/startup-skill.git .agents/startup-skill

Then reference the skill from .agents/startup-skill/startup-design/.

Option 5: Fork and Customize

1. Fork this repository
2. Customize the skill for your specific needs
3. Clone your fork into your projects

Option 6: SkillKit (Multi-Agent)

Use SkillKit to install skills across multiple AI agents (Claude Code, Cursor, Copilot, etc.):

# Install the skill (auto-detects Claude Code, Cursor, Copilot, etc.)
npx skillkit install ferdinandobons/startup-skill

Usage

Start a conversation with your AI agent and describe your startup idea:

I have an idea for a SaaS that helps real estate agents automate their follow-up emails using AI.
I've been an agent for 12 years and I know the pain. Can you help me figure out if it's worth building?

The skill will automatically trigger and guide you through the intake interview, then proceed phase by phase.

To resume a previous session:

Resume from checkpoint

For a quick validation:

I need a quick validation of this idea: [your idea]. Fast track mode.

Repository Structure

startup-skill/
├── startup-design/
│   ├── SKILL.md                          # Main skill file (8 phases)
│   └── references/
│       ├── research-principles.md        # Cross-cutting research rules
│       ├── research-wave-1-market.md     # Market sizing agents
│       ├── research-wave-2-competitors.md # Competitive analysis agents
│       ├── research-wave-3-customers.md  # Customer research agents
│       ├── research-wave-4-distribution.md # Distribution agents
│       ├── research-synthesis.md         # How to synthesize findings
│       └── frameworks.md                 # Strategic frameworks
├── CLAUDE.md                             # Agent guidelines
├── CONTRIBUTING.md                       # How to contribute
├── LICENSE                               # MIT License
└── README.md                             # This file

Eval Results

The skill has been tested across 5 eval scenarios comparing skill-guided output vs. baseline:

Metric	With Skill	Without Skill
Assertion Pass Rate	100% (21/21)	57% (12/21)
Structured Output	Always	Rarely
Interview Process	Multi-round, targeted	Single dump
Brainstorming	5-8 variations	Skipped
Progress Tracking	Always	Never

Contributing

See CONTRIBUTING.md for details on how to contribute.

Quick Start

1. Fork the repository
2. Create a feature branch (git checkout -b feature/improvement-name)
3. Make your changes
4. Test with an AI agent
5. Submit a pull request

License

MIT