dvcrn When to Use
User wants new business ideas, startup concepts, or side project inspiration. Agent generates ideas using proven frameworks, filters by constraints, and validates viability.
Architecture
Memory lives in ~/business-ideas/. See memory-template.md for setup.
~/business-ideas/
├── ideas.md # HOT: generated ideas with scores
├── favorites.md # WARM: ideas user marked for exploration
├── filters.md # User's default filters and preferences
└── archive/ # COLD: rejected or explored ideas
Quick Reference
| Topic | File |
|---|---|
| Memory setup | memory-template.md |
| Generation frameworks | frameworks.md |
| Validation methods | validation.md |
Core Rules
1. Never Repeat Ideas
Before generating, scan ~/business-ideas/ideas.md for similar concepts. Each idea must be meaningfully different from previous generations.
2. Always Apply Filters
Ask for or use stored filters before generating:
| Filter | Options |
|---|---|
| Industry | Tech, Health, Finance, Education, Consumer, B2B, Creator |
| Model | SaaS, Marketplace, Agency, Product, Content, Service |
| Investment | Bootstrap ($0-1K), Seed ($1K-50K), Funded ($50K+) |
| Time | Side project (5h/week), Part-time, Full-time |
| Skills | Technical, Non-technical, Hybrid |
3. Score Every Idea
Rate each idea on 5 dimensions (1-10):
| Dimension | Question |
|---|---|
| Market | Is there proven demand? |
| Timing | Why now? What changed? |
| Moat | Can this be defended? |
| Founder-fit | Does user have unfair advantage? |
| Simplicity | Can MVP ship in 30 days? |
Viability = average score. Flag ideas scoring 7+ as high-potential.
4. Use Frameworks Systematically
Rotate through frameworks to ensure variety. See frameworks.md for complete list:
- Pain Point Mining
- Trend Riding
- Existing Business Remix
- Audience First
- Technology Arbitrage
5. Batch Generation Mode
When user asks for "ideas" (plural) or "brainstorm":
- Generate 5-10 ideas minimum
- Use multiple frameworks
- Include mix of safe bets and moonshots
- Present as ranked table
6. Deep Dive on Request
When user picks an idea to explore:
- Expand business model canvas
- Identify 3 biggest risks
- Suggest validation experiments
- Estimate time-to-revenue
- Save to favorites
7. Update Memory Proactively
| Event | Action |
|---|---|
| Ideas generated | Append to ideas.md with date |
| User likes idea | Move to favorites |
| User rejects idea | Note rejection reason |
| User sets preference | Update filters.md |
Idea Generation Traps
- Generic ideas without specificity → Always include target customer and unique angle
- Ideas requiring massive scale → Include bootstrappable alternatives
- Pure tech plays without business model → Define revenue from day one
- Copying without differentiation → Require "10x better" or "10x cheaper" angle
- Ignoring user constraints → Always check filters first
Related Skills
Install with clawhub install <slug> if user confirms:
business— Strategy and planningstartup— Launch and scaleindie-hacker— Bootstrap and grow solo
Feedback
- If useful:
clawhub star business-ideas - Stay updated:
clawhub sync